A.-Y. Wang1, H.-Y. Hu1, Y.-N. Ou1, Z.-T. Wang1, Y.-H. Ma1, L. Tan1, J.-T. Yu1,2,3
1. Department of Neurology, Qingdao Municipal Hospital, Qingdao University, Qingdao, China; 2. Department of Neurology and National Center for Neurological Disorders, Huashan Hospital, State Key Laboratory of Medical Neurobiology and MOE Frontiers Center for Brain Science, Shanghai Medical College, Fudan University, Shanghai, China; 3. The Institute of Science and Technology for Brain-inspired Intelligence, Fudan University, Shanghai, China.
Corresponding Author: Prof. Jin-Tai Yu (ORCID 0000-0002-3310-5875), National Center for Neurological Disorders, Huashan Hospital, Shanghai Medical College, Fudan University, 12th Wulumuqi Zhong Road, Shanghai 200040, China, email@example.com; or Prof. Lan Tan (ORCID 0000-0002-8759-7588), Department of Neurology, Qingdao Municipal Hospital, Qingdao University, No.5 Donghai Middle Road, Qingdao, China, firstname.lastname@example.org, Tel: +86 21 52888163; Fax: +86 21 62483421.
J Prev Alz Dis 2022;
Published online September 27, 2022, http://dx.doi.org/10.14283/jpad.2022.81
Background: In recent decades, increased attention has been paid to the impact of socioeconomic status (SES) on cognition function and dementia, however, an ongoing debate continues to exist. The objective of our study was to explore the potential effect of SES on the risks of cognitive dysfunction and dementia.
Methods: PubMed, Cochrane Library, and EMBASE were searched for prospective studies from inception to 9 January 2022. Meta-analyses using random-effect models were performed, and then subgroup analyses stratified by study characteristics for specific outcomes were conducted.
Results: Thirty-nine prospective studies (1,485,702 individuals) were eligible for inclusion, of which 25 reported the incidence of dementia and 14 reported cognitive decline. Primary results of the meta-analyses found an elevated combined risk of cognitive impairment and dementia (relative risk [RR] = 1.31, 95% confidence interval [CI] = 1.16-1.49) in low-SES participants compared with high-SES participants. We also found an elevated risk of all-cause dementia (RR = 1.40, 95% CI = 1.12-1.74) in low-SES participants. Further subgroup analyses stratified by education, occupation, and income showed that low education subgroup (RR = 1.21, 95% CI = 1.04-1.41) and low-income subgroup (RR = 1.22, 95% CI = 1.10-1.35) had an increased combined risks of cognitive impairment and dementia, but only individuals with lower education had a higher risk of dementia (RR = 1.66, 95% CI = 1.20-2.32).
Conclusions: Low SES substantially increased the risk of dementia and cognitive dysfunction, suggesting that public health strategies could reduce the dementia burden by reducing social inequalities.
Key words: Dementia, cognition, socioeconomic status, meta-analysis.
As a syndrome characterized by severe cognitive decline, dementia had a rapidly increasing incidence (1, 2). It imposed not just a large health burden but also a heavy economic burden on individuals, families, and societies (3), costing an estimated US$ 2 trillion globally for dementia by 2030 (4). Owing to the absence of curative strategies, primary prevention, which aims to delay the onset and therefore reduce the incidence of dementia, is particularly important (5). Hence, figuring out what variables are related to dementia has become incredibly pressing. Several factors have been demonstrated to be related to the development of dementia and cognitive declines, such as environmental factors (6), education (7), multi-gene (8), the complexity of work (9), and income (10).
Socioeconomic status (SES) is a complex concept reflecting a person’s overall status in society, which encompasses income, education, and occupation (11). Recently, SES is no longer simply regarded as a confounding factor but as a research topic of interest. Additionally, several studies have reported that low SES is associated with many adverse health outcomes (11, 12). Previous cross-sectional studies indicated that socioeconomic status might influence individual cognitive function (13), but those studies are limited by cross-sectional design which limits causal inferences. Moreover, numerous cohort studies investigated the association between SES and cognitive function. However, estimates of the longitudinal relationship between SES and cognitive decline or dementia differed broadly across studies. By comparing intracranial volume, Walhovd demonstrated that SES has no uniform association with brain and cognition (14). Instead, Hazzouri et al have an opposite view (15). The sources of heterogeneity could be the population heterogeneity, the lack of uniform SES standards, and the small sample sizes.
Some reviews and meta-analyses have described the effects of socioeconomic factors on dementia. Xu et al. found that each additional year of education was associated with a 7% lower risk of dementia (7). In the studies by Huang et al., mental work or more complex work were also associated with a lower risk of dementia (16). Establishing a policy of minimum guaranteed livable income would reduce the incidence of dementia in Andrew’s study (17). However, they analyzed only part of the SES indicators. Canevelli et al. noticed SES has vital research and clinical relevance for AD, but they were not sufficient samples to make any inferences on the relationship between SES and the risk of AD (18). Besides, low SES, as one of the environmental factors, was considered a key risk factor for cognition dysfunction in Zhao and Wu’s study (6, 19). Nevertheless, they lacked a standardized definition of SES and a comprehensive analysis of the association between SES and cognition dysfunction. To obviate the above problems, we conducted a systematic review and meta-analysis to summarize the association between SES and dementia or cognitive decline based on the published prospective research data. Moreover, we conducted subgroup analyses stratified by basic population characteristics, and components of SES factors that might cause heterogeneity, separately.
Search strategy and selection criteria
The PRISMA 2020 guideline was followed to conduct this systematic review and meta-analysis (20) (Supplementary Appendix 1). The protocol has been registered with PROSPERO, number CRD42022302750. Three databases, including PubMed, Cochrane Library, and EMBASE, were searched from inception until January 9, 2022. We searched three databases using the following search terms: (“dementia” OR “Alzheimer” OR “Alzheimer’s” OR “ACD” OR “neurodegeneration” OR “neurodegenerative” OR “cognitive” OR “cognition”) AND (“Socioeconomic” OR “Socio-economic” OR “sociodemographic” OR “socio-demographic” OR “socioecological” OR “socio-ecologic” OR “living Standard” OR “Income” OR “Economic” OR “Social” OR “Education” OR “educational” OR “indigents” OR “Poverty” OR “Employment” OR “Family Characteristic” OR “occupation” OR “wage” OR “salary” OR “TDI” OR “SES” OR “deprivation”) AND (“cohort” OR “longitudinal” OR “prospective” OR “follow-up”). No restrictions were added. In addition, bibliographies of identified studies and relevant systematic reviews and meta-analyses were scrutinized for potentially omitted studies.
In order to qualify for analysis, studies had to meet six criteria: 1 ) utilized a longitudinal prospective cohort or nested case-control study design; 2 ) explored the association of SES (income, education, and occupation) with the risk of cognitive decline or dementia; 3 ) was published in English; 4 ) either reported risk estimates (adjusted relative risk [RR], hazard ratio [HR], or odds ratio [OR]), with its corresponding 95% confidence interval (CI) or sufficient data to calculate these; 5) the population was diagnosed without dementia at baseline and was followed up over 1 year; 6) the outcome was dichotomous variables. Studies would be excluded if they 1) did to meet the inclusion criteria, 2) were not original research (reports, letters, commentaries, conference abstracts, reviews, meta-analyses, editorials) (21).
In each included article, the following information and data were extracted: year of publication, title, first author, region, mean age at baseline, the number of participants included, gender (female %), follow-up time, outcomes, diagnostic criteria for dementia and cognitive decline, SES measures (comprehensive SES /income/education/occupation), multivariable-adjusted effect sizes (RR, OR, HR, 95% CI) and confounders adjusted for. When multiple articles reported data from the same cohort, the articles that included the largest number of participants, or published in the latest year were included. But if articles from the same cohort evaluated different aspects of SES or had different cognitive outcomes, these articles were separately included in different analyses. If a publication reported the crude and multivariable-adjusted risk estimates concurrently, we selected the risk estimate in the model adjusted for most covariates. The hazard ratio reported in the eligible studies was considered a proxy for RR. For studies which only reported OR but not RR or HR, we calculated RR using the following algorithm: RR adjusted = OR adjusted / [(1 − P0) + (P0 × OR adjusted)]. P0 represents the incidence of cognitive decline or dementia in the non-exposed group (22). In some studies where P0 was not available, the overall incidence of outcomes was used as a proxy. When there were multiple SES indicators in a study, we selected the most comprehensive and the most SES-related indicators in the main analysis, and other indicators were additionally extracted for subgroup analyses. Two independent investigators examined the research results. When investigators provided controversies, disagreements were resolved via discussion.
Assessment of study quality
The Newcastle-Ottawa Quality Assessment Scale (NOS) was selected as a tool for independent quality assessments by two investigators (23). The higher the score, the better the quality of the studies, with 9 points being the maximum score. Articles with a score lower than 5 are regarded to be of bad quality. No article was excluded due to a low score. To ensure the reliability of our results, we also combined articles with NOS scores ≥7 or articles with NOS scores ≥ 8, respectively.
Heterogeneity was quantiﬁed using I² values, where I² ≥ 60% indicated substantial heterogeneity and I² ≥ 75% indicated considerable heterogeneity. Since the heterogeneity was great across different studies (I2 > 60%), the pooled effect was analyzed using a random effect model. We took the two most extreme levels for comparison (the lowest quintile versus the highest quintile) (24), with the highest SES index (including highest education, the highest income, and highest occupation attainment) set as the reference group. If the article used the lowest SES index as the reference group, we carried out a corresponding conversion (25). Egger’s regression test was applied to assess potential publication bias. If publication bias was presented, we applied a trim-and-fill method to evaluate potential missing data and reassessed the aggregate results. Additionally, we conducted sensitivity analyses to test whether the results were stable. Further, to find the possible source of heterogeneity, meta-regression analyses were performed on region, age (mean), sex (female %), different adjusted factors (whether the study adjusted for age and SES indicators), follow-up time, baseline cognition, response rate, SES measure, and so on. Finally, subgroup analyses were carried out according to baseline cognition (non-dementia vs. cognitive normal), mean age (< 65 years vs. ≥ 65 years), SES measure (comprehensive SES / income/education/occupation), NOS Scale score (< 7 vs. ≥ 7), follow-up time (< 8 years vs. ≥ 8 years), and response rate (> 70% vs. ≤ 70%), gender (female≥50% vs. females <50%). Since measurement errors are prone to occur in studies applying self-questionnaires, we also combined data but excluded self-questionnaire studies. All analyses in our study were performed using R 4.1.0 and Stata SE16.0 software, and a two-sided P<0.05 was considered statistically significant.
Figure 1 showed the detailed screening procedure, a total of 113,630 articles were identified from PubMed, Cochrane Library, and EMBASE databases, and 82,190 articles remained after de-duplication. Then, 223 were identified for full-text assessment after an initial screening of titles and abstracts of these studies, a total of 39 articles were included after screening full texts (202 articles) and citations.
1. Repetitive cohort (n=8); No comparison of cognitive level (n=4); No comparison of socioeconomic status(n=9); Related to Parkinson’s disease(n=4); SES as covariate only(n=2)
A total of 39 articles (38 prospective cohort studies; 1 prospective case-control study (26)) with 1,485,702 individuals (minimum sample size n =321; median n = 45,021; maximum n = 1,341,842) were included. The basic characteristics of included studies are presented in Table 1. Fourteen studies were for cognitive decline (24, 27-39), and 25 studies were for dementia (26, 35, 40-62). See Supplementary Appendix 3 for details criteria of cognitive impairment. Among the 39 included studies, 33 studies from independent cohorts were included in the primary analysis, and 6 studies from duplicate cohorts (31-36) were only for specific sub-analyses. Notably, 5 independent cohorts reported Alzheimer’s disease (AD) (40, 51, 55, 59, 62) as the outcome, and 16 independent cohorts reported all-cause dementia (ACD) as the outcome.
* Article included in occupation sub-analysis; † Article included in education sub-analysis; Abbreviations: SES: socioeconomic state; SEP: socioeconomic position; NA, not accessible; RR, risk ratio; CI, confidence interval; IQCODE, The Informant Questionnaire on Cognitive Decline in the Elderly screening; CIND, cognitive impairment no-dementia ;SPMSQ, The Short Portable Mental Status Questionnaire ;CDR, The Clinical Dementia Rating; FAQ, Functional Assessment Questionnaire; NIA-AA, criteria of the National Institute on Aging and the Alzheimer’s Association; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th ed.; MMSE, Mini-mental State Examination; NINDS-AIREN, the National Institute of Neurological Disorders and Stroke-the Association In-ternationale pour la Recherche et 1’Enseignement en Neurosciences diagnostic criteria; NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association diagnostic criteria; ICD, international Classi-fication of diseases; NPR and CDR, the National Inpatient Register and the Cause of Death Register using International Classification of Dis-eases codes; CDR, the global Clinical Dementia Rating; LCT, the Leganés Cognitive Test; 3MSE, the Modified Mini-Mental State Examina-tion; Kungsholmen, the Kungsholmen Project; PAQUID, Personnes Agées QUID cohort; SALSA, the Sacramento Area Latino Study on Ag-ing cohort; CLHLS, the Chinese Longitudinal Healthy Longevity Survey; ELSA, the English Longitudinal Study of Ageing; KLoSAa, the Korean Longitudinal Study of Ageing (2006–2019); InCHIANTI: the population-based aging in the Chianti Area; ILSE, the Interdisciplinary Longitudinal Study of Adult Development and Aging; ZARADEMP: the zaragoza dementia and depression Project; Whitehall II, the White-hall II study; WHS, the Women’s Health Study; ACT: the Adult Changes in Thought; NIAE, the National Institute on Aging Established; KYS, the Korean Yonchun Survey; CDR, the Cause of Death Register ; BRAiNS, (the University of Kentucky AD Center); IMIAS, the In-ternational Mobility in Aging Study; IIHD, the Israeli Ischemic Heart Disease Project; Health ABC, the Health, Aging and Body Composition; NSJE, National Survey of the Japanese Elderly; CAIDE, the Cardiovascular Risk Factors, Aging and Dementia; JAGES , the Japan Geronto-logical Evaluation Study; NHS, the Nurses’ Health Study; NHAT ,the National Health and Aging Trends Study; Vallecas, the Vallecas Project cohort; ISA, the Ibadan Study of Aging; CREDOS, the Clinical Research Center for Dementia of South Korea
Moreover, the 39 included articles applied different SES designs and criteria, among which 10 articles had a comprehensive indicator of SES (24, 28, 31, 37, 40, 44-46, 52, 59), as detailed in Supplementary Appendix 2, while the other articles only separately described one or two aspects of SES measures (income, education, and occupation). The degrees of factor adjustment of all articles can be found in Supplementary Appendix 4. The average NOS score for all 39 articles was 7.35, and the details are presented in Supplementary Appendix 5.
Primary results of meta-analyses
First, we analyzed the association of SES with the combined risk of cognitive impairment and dementia (n = 33). Since the studies were heterogeneous, we used a random-effects model in all effect size calculations. The data showed that lower SES played an important role in increasing the combined risk of cognitive impairment and dementia, with substantial heterogeneity (RR = 1.31, 95% CI = 1.16-1.49, I² =73%; Figure 2) and no publication bias (P-value = 0.403, Egger’s- test). Further, fourteen studies were included in the analysis of cognitive impairment and the pooled RR was 1.27 (95% CI = 1.13-1.42, I² = 97%; Figure 2) and this positive relationship was also evident for ACD (RR = 1.40, 95% CI = 1.12-1.74, I²=80%; Figure 2). Also, no publication bias was detected (P=0.242 for cognitive impairment; P=0.521 for ACD) by using Egger’s test. Notably, when we included only five AD articles for the meta-analysis, no obvious association between SES and AD was found (RR = 1.19, 95% CI = 0.70-2.03, I²=83%; Figure 2). A chart of the forest can be seen in Supplementary Appendix 6 and 7.
Low SES was associated with an increased combined risk of cognitive impairment and dementia, and this positive relationship was also evident for cognitive impairment and ACD. No significant association was found between SES and AD. Heterogeneity was high in all the above analyses. Abbreviations: SES, socioeconomic state; RR, relative risk; CI, confidence interval; ACD, all-cause dementia; AD, Alzheimer’s disease.
Sensitivity analyses for cognitive impairment/dementia (both cognitive impairment and dementia), cognitive impairment, ACD, and AD were performed. Sensitivity analyses showed that the results were robust after removing each article separately and merging the effect values of the remaining articles (Supplementary Appendix 8 and 17). When we have excluded 3 cohorts that used the self-reported questionnaires, results showed that lower SES could increase 34% risk of cognitive impairment and dementia (95% CI =1.17-1.55).
Results of meta-regression and subgroup analyses
Meta-regression on RR of cognitive impairment and dementia according to region, outcome, response rate, age (mean), follow-up years, baseline cognition, adjusted-factors, SES measure, and NOS score. Differences in SES measures were found in the analyses (P =0.020; Supplementary Appendix 9). Therefore, we carried out a separate meta-analysis on different SES indications. Statistically significant positive relationship existed in participants with lower education (RR = 1.22, 95% CI = 1.10-1.35, I²=96%; n = 20 cohort; 1,464,091 participants; Figure 3), those with lower comprehensive SES (RR = 1.75, 95% CI = 1.37-2.23, I²=65%; n=10 studies; 24,043 participants; Figure 3) and those with lower income (RR = 1.21, 95% CI = 1.04-1.41, I²=70%; n = 11 studies; 91,301 participants; Figure 3). However, when we conducted a meta-analysis on the association of occupation with the combined risk of cognitive impairment and dementia, no relationship was observed (RR = 1.06, 95% CI = 0.83-1.36, high heterogeneity, n=16 cohort; 112,000 participants, Figure 3). In addition, the association was still significant in subgroup analyses according to baseline cognition, mean age, NOS Scale score, response rate. More details can be found in Figure 3 and Supplementary appendix 13. Significant differences were observed in the sex subgroups (RR = 1.27, 95% CI = 1.10-1.45, 26 studies for female ≥ 50%; RR = 1.60, 95% CI = 0.97-2.64, 6 studies for females < 50%; Supplementary appendix 25). There is a trend that individuals with longer follow-up time were more likely to have cognitive dysfunction (RR = 1.22, 95% CI = 1.04-1.43, I² =71%, 20 studies for follow-up time < 8 years; RR = 1.46, 95% CI = 1.24-1.72, I² =67%, 13 studies for follow-up time ≥ 8 years; Figure 3). The forest plot (Supplementary Appendix 10-15, 25) shows the above results.
The association of low SES with the combined risk of cognitive impairment and dementia were significant in subgroup analyses according to baseline cognition, SES measures, mean age, NOS Scale score, follow-up time, and response rate. But no significant association was found between occupation and the risk of cognitive impairment and dementia. The heterogeneity was reduced with subgroups analyzed, but evident heterogeneity existed in the majority of subgroups analyzed. Abbreviations: RR, relative risk; CI, confidence interval.
We next conducted a meta-regression of ACD according to region, study years, response rate, age, follow-up time, baseline cognition, adjusted factors, SES measure, and NOS score. The results were all non-significant (Supplementary Appendix 16). In addition, subgroup analyses according to baseline cognition, mean age, SES measure, NOS scale score, follow-up time, and response rate showed that the association of SES with ACD remained significant in subgroups of non-demented participants at baseline, follow-up time ≥ 8, mean age ≥ 65, low SES and low education. More detailed information can be seen in Figure 4. The association between SES and ACD was not significant, when baseline cognition was normal (RR = 1.07, 95% CI = 0.45-2.55, I² =0%, 2 studies; Figure 4), the mean age of the study population was less than 65 years (RR = 1.47, 95% CI = 0.52-4.16, I² =88%, 3 studies; Figure 4), the follow time was less than 8 years (RR = 1.29, 95% CI = 0.97-1.72, I² =76%, 8 studies; Figure 4), SES measure were income and occupation (RR = 1.19, 95% CI = 0.78-1.82, I² =77%, 4 studies for income; RR = 1.03, 95% CI = 0.77-1.36, I² =68%, 4 studies for occupation; Figure 4), which should be interpreted with caution, since only 2 studies of participants with normal baseline cognition and 3 studies of participants with mean age < 65 years. Non-significant differences in other subgroups can be seen in Figure 4. Supplementary Appendix 19-24 depicts forest plots for the above results.
The association of SES with ACD remained significant in subgroups of non-demented participants at baseline, mean age ≥65, follow-up time ≥ 8, low SES and low education. No differences were observed among the NOS scale score and response rate subgroups. Heterogeneity was high across the majority of analyses. Abbreviations: RR, relative risk; CI, confidence interval.
Based on the meta-analysis of 39 prospective studies, we found that lower SES could increase 31% risk of cognitive impairment and dementia and increase 40% risk of ACD. Specifically, low levels of education and income (two main SES indicators) were found to be substantially related to an elevated combined risk of cognitive impairment and dementia, but only low education was a risk factor for dementia. Our study provides insights into the connections between SES measures and cognitive function and has potentially important implications for health policy.
Our main findings were consistent with most previous systematic reviews and meta-analyses. Canevelli Bruno’s (18) systematic review of 49 studies showed the significant impact of SES indicators on AD. Compared with Canevelli Bruno’s study, we used meta-analysis to quantify the relationship between SES and dementia. A previous review (63) incorporating 65 articles on the Chinese population and 29 articles on the Indian population also found that low levels of education and income were related to increased risk of dementia. Mika (17) analyzed two Finnish cohort studies and reported that low SES was associated with increased risk for a range of 18 diseases including dementia. With broader population sources, we observed that low SES increased not only the risk of dementia but also the risk of cognitive impairment. Korous et al., in a recent meta-analysis (64), found that higher SES is related to more favorable cognitive ability compared to lower SES. With more included studies, we investigated SES was associated with cognitive impairment and dementia. Besides, we also tested if age, follow-up time, the quality of articles, and response rate could affect our results by performing multiple subgroup analyses. Lower education, as a key indicator of SES, has been demonstrated to be a risk factor in many previous reviews (65). And our result was consistent with previous reviews. However, two recent systematic reviews and meta-analysis found that mental work or more complex work were also associated with a lower risk of dementia. These results differ from our result, which may be attributed to a difference in different classification standards for work (16).
The exact mechanisms of SES affecting cognition are less well understood, involving the social or psychological aspect and the anatomic or physiologic aspect. SES has been found to be associated with a high risk of obesity, sedentarism, diabetes, hypertension, cardiovascular disease, metabolic syndrome, stroke, and mental diseases (66), all of which also raise the risk of dementia. At the social or psychological levels, firstly, some assumed that high income subtly reduces the incidence of dementia via promoting healthier lifestyles (67), such as a healthier diet, moderate drinking, regular physical activity, as well as blood pressure, cholesterol, or diabetes management. Secondly, some assumed that highly educated people can make better use of cognitive reserves and reduces the risk of dementia by increasing the efficiency of cognitive reserves (68) and cognitive networks among elderly people (69). Thirdly, individuals with high SES often were exposed to more cognitively stimulating environments (70) and less social discrimination (71), which might play a key role in brain development or activation and therefore reduce the risk of cognitive decline (41, 72). Fourthly, individuals with low SES are more likely to be socially isolated as they experience less social contact with others. Evans et al. found that low levels of social isolation were associated with better cognitive function (73).
At the anatomic or physiologic level, recent work (74) suggested that psychosocial stress could be an intermediate of the relationship between SES and dementia. Long-term psychosocial stress can cause physiologic changes in the hypothalamic-pituitary-adrenal (HPA) axis and decrease HPA hormone receptor density in the hippocampus, further weakening brain functions and predisposing individuals to dementia (75). Besides, it has been suggested that education may influence cognitive function by increasing synaptic density (68, 76). When dementia occurs, neurons and synapses will be disrupted, and the presence of increased synaptic density early in life delays the occurrence of cognitive decline and other clinical symptoms late in life (72). Furthermore, brain structure also varied by SES. Several previous articles have demonstrated that SES was associated with reduced cortical surface area, cortical tissue volume, and cortical thickness (77), in particular the hippocampus and frontal cortex (78, 79). This structural change in the brain may result in differences in executive and memory functions.
It is worth noting that the relationship between SES and cognition can be explained using health economics (80). Grossman’s health demand model serves as a foundation for health economics theory. In Grossman’s model, health is viewed as a form of capital that can be invested in (81). People with a higher level of education are more likely to invest in their health. The higher the income, the higher the ability to pay for health care services and the higher the effective demand for health care services (82). Those with higher SES are more likely to live and work away from bad physical environments (83). Therefore, those with high SES would have better access to health care and better disease prevention, which further reduce the risk of cognitive decline or dementia.
In our subgroup analyses for the combined risk of cognitive dysfunction and dementia, there was no significant difference between subgroups (except for sex subgroup). In low SES, males are more likely to develop cognitive impairment/dementia than females. Interestingly, in our subgroup analyses for ACD risk, individuals with high SES in the subgroup of follow-up ≥ 8 years or mean age < 65 years old showed a lower risk of developing ACD, but the association became non-significant in the subgroups of follow-up < 8 years or mean age < 65 years old. A possible reason for this discrepancy is as follows. Since the development of dementia is a long process, dementia cannot be well observed when participants are too young, or when the follow-up time is too short. Moreover, no association between low SES and ACD in the subgroup of cognitively normal participants at baseline was found, which can be explained by the limited sample size of this subgroup and by the fact that cognitively normal participants were less prone to dementia than those with mild cognitive impairment. Additionally, there was no obvious relationship between AD and SES in our meta-analysis. One possible reason is that the data available are not suﬃcient to assess the relationship of SES with the risk of AD and the number of articles we included was limited. Another possible reason is the incidence of AD is low and not easily diagnosed compared to ACD or cognitive decline. When we have excluded 3 cohorts that used the self-reported questionnaires, we found that the impact of SES on cognition becomes more pronounced. Insufficient accuracy of self-questionnaire may be the reason for this result.
Our study has several strengths. First, our main advantage of this article is that we only included prospective cohort studies, which minimize memory and choice biases compared to retrospective or bidirectional studies. Second, our combined effect value is RR, which has more direct clinical significance than other effect values. In addition, developing dementia is a gradual course, and a longer follow-up period will allow us to have more accurate results. Thus, we had an average follow-up period of ten years. Notwithstanding, our study also has many inevitable limitations. Firstly, our meta-analysis showed great heterogeneity across the included studies, and we performed meta-regression and subgroup analyses to reduce heterogeneity. Secondly, due to limited data, we did not classify the types of dementia. Thirdly, there is no unified standard for cognitive impairment currently. However, we have collated the specific standards of cognitive impairment in the Supplementary Appendix. Fourthly, the observational studies included in this study did not demonstrate a causal relationship, and more studies allowing for causal inference are needed in the future, for example, mendelian randomization studies and studies on the relationship between SES and brain pathological changes in normal people. Fifthly, health investments and social isolation are associated with both SES and cognitive decline, and health investment and social isolation should be important observation factors in future articles. Sixthly, since the components of SES may influence each other, more comprehensive SES criteria are needed in the future.
In conclusion, our findings provided a piece of evidence that low SES was linked to an increased risk of dementia and cognitive dysfunction, with both education and income showing a high association. But the association between work and dementia needs to be further explored. Our study research indicated that socioeconomically disadvantaged individuals were more likely to experience adverse health effects. Thus, public health strategies for dementia prevention should target socioeconomic inequalities to reduce health disparities, and policy-makers should develop effective public health and reduce socioeconomic disparities to mitigate the possible poor health outcomes (31). According to our results, reducing socioeconomic inequality should first attention to educational inequality and income inequality, especially in developed countries. Global public health strategies should currently aim toward universal education coverage and income redistribution. However, there is still a long way to go in the achievement of socioeconomic equality.
Acknowledgments: This study was supported by grants from the National Natural Science Foundation of China (82071201, 81971032), Shanghai Municipal Science and Technology Major Project (No.2018SHZDZX01), Research Start-up Fund of Huashan Hospital (2022QD002), Excellence 2025 Talent Cultivation Program at Fudan University (3030277001), and ZHANGJIANG LAB, Tianqiao and Chrissy Chen Institute, and the State Key Laboratory of Neurobiology and Frontiers Center for Brain Science of Ministry of Education, Fudan University.
Conflicting interests: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Authors’ contributions: LT and JTY are responsible for the study conception and design. AYW, HYH, YNO, ZTW and YHM conducted the study. AYW, HYH, and YNO analyzed and extracted data. AYW, HYH, and JTY wrote the first draft of the manuscript. All authors have contributed to the manuscript by revising and editing critically for important intellectual content and given final approval of the version and agreed to be accountable for all aspects of the work presented here.
Ethical Standards: None.
1. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022.Doi: https://10.1016/s2468-2667(21)00249-8
2. Prince M, Bryce R, Albanese E, et al. The global prevalence of dementia: a systematic review and metaanalysis. Alzheimers Dement. 2013;9(1):63-75.e2.Doi: https://10.1016/j.jalz.2012.11.007
3. 2016 Alzheimer’s disease facts and figures. Alzheimers Dement. 2016;12(4):459-509.Doi: https://10.1016/j.jalz.2016.03.001
4. The Lancet N. Response to the growing dementia burden must be faster. Lancet Neurol. 2018;17(8):651.Doi: https://10.1016/s1474-4422(18)30256-4
5. Norton S, Matthews FE, Barnes DE, Yaffe K, Brayne C. Potential for primary prevention of Alzheimer’s disease: an analysis of population-based data. Lancet Neurol. 2014;13(8):788-94.Doi: https://10.1016/s1474-4422(14)70136-x
6. Zhao YL, Qu Y, Ou YN, et al. Environmental factors and risks of cognitive impairment and dementia: A systematic review and meta-analysis. Ageing Res Rev. 2021;72:101504.Doi: https://10.1016/j.arr.2021.101504
7. Xu W, Tan L, Wang HF, et al. Education and Risk of Dementia: Dose-Response Meta-Analysis of Prospective Cohort Studies. Mol Neurobiol. 2016;53(5):3113-23.Doi: https://10.1007/s12035-015-9211-5
8. Jansen IE, Savage JE, Watanabe K, et al. Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer’s disease risk. Nat Genet. 2019;51(3):404-13.Doi: https://10.1038/s41588-018-0311-9
9. Hyun J, Hall CB, Katz MJ, et al. Education, Occupational Complexity, and Incident Dementia: A COSMIC Collaborative Cohort Study. Journal of Alzheimer’s disease : JAD. 2021.Doi: https://10.3233/JAD-210627
10. Roehr S, Pabst A, Luck T, Riedel-Heller SG. Is dementia incidence declining in high-income countries? A systematic review and meta-analysis. Clin Epidemiol. 2018;10:1233-47.Doi: https://10.2147/clep.S163649
11. Russ TC, Stamatakis E, Hamer M, et al. Socioeconomic status as a risk factor for dementia death: individual participant meta-analysis of 86 508 men and women from the UK. Br J Psychiatry. 2013;203(1):10-7.Doi: https://10.1192/bjp.bp.112.119479
12. Kivimäki M, Batty GD, Pentti J, et al. Association between socioeconomic status and the development of mental and physical health conditions in adulthood: a multi-cohort study. Lancet Public Health. 2020;5(3):e140-e9.Doi: https://10.1016/s2468-2667(19)30248-8
13. Shih RA, Ghosh-Dastidar B, Margolis KL, et al. Neighborhood socioeconomic status and cognitive function in women. Am J Public Health. 2011;101(9):1721-8.Doi: https://10.2105/ajph.2011.300169
14. Walhovd KB, Fjell AM, Wang Y, et al. Education and Income Show Heterogeneous Relationships to Lifespan Brain and Cognitive Differences Across European and US Cohorts. Cerebral cortex (New York, NY : 1991). 2021.Doi: https://10.1093/cercor/bhab248
15. Zeki Al Hazzouri A, Haan MN, Osypuk T, et al. Neighborhood socioeconomic context and cognitive decline among older Mexican Americans: results from the Sacramento Area Latino Study on Aging. Am J Epidemiol. 2011;174(4):423-31.Doi: https://10.1093/aje/kwr095
16. Huang LY, Hu HY, Wang ZT, et al. Association of Occupational Factors and Dementia or Cognitive Impairment: A Systematic Review and Meta-Analysis. J Alzheimers Dis. 2020;78(1):217-27.Doi: https://10.3233/jad-200605
17. Andrew MK. Midlife improvements in financial situation and dementia risk: Considerations for research, policy, and practice. International Psychogeriatrics. 2020;32(11):1271-3.Doi: https://10.1017/S1041610219002035
18. Canevelli M, Bruno G, Vico C, et al. Socioeconomic disparities in clinical trials on Alzheimer’s disease: a systematic review. Eur J Neurol. 2018;25(4):626-e43.Doi: https://10.1111/ene.13587
19. Wu YT, Prina AM, Brayne C. The association between community environment and cognitive function: a systematic review. Soc Psychiatry Psychiatr Epidemiol. 2015;50(3):351-62.Doi: https://10.1007/s00127-014-0945-6
20. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Bmj. 2021;372:n71.Doi: https://10.1136/bmj.n71
21. Xu W, Tan CC, Zou JJ, Cao XP, Tan L. Sleep problems and risk of all-cause cognitive decline or dementia: an updated systematic review and meta-analysis. J Neurol Neurosurg Psychiatry. 2020;91(3):236-44.Doi: https://10.1136/jnnp-2019-321896
22. Grant RL. Converting an odds ratio to a range of plausible relative risks for better communication of research findings. Bmj. 2014;348:f7450.Doi: https://10.1136/bmj.f7450
23. Stang A. Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol. 2010;25(9):603-5.Doi: https://10.1007/s10654-010-9491-z
24. Fernández-Blázquez MA, Noriega-Ruiz B, Ávila-Villanueva M, et al. Impact of individual and neighborhood dimensions of socioeconomic status on the prevalence of mild cognitive impairment over seven-year follow-up. Aging Ment Health. 2021;25(5):814-23.Doi: https://10.1080/13607863.2020.1725803
25. Orsini N. From floated to conventional confidence intervals for the relative risks based on published dose-response data. Comput Methods Programs Biomed. 2010;98(1):90-3.Doi: https://10.1016/j.cmpb.2009.11.005
26. Hendrie HC, Smith-Gamble V, Lane KA, et al. The Association of Early Life Factors and Declining Incidence Rates of Dementia in an Elderly Population of African Americans. J Gerontol B Psychol Sci Soc Sci. 2018;73(suppl_1):S82-s9.Doi: https://10.1093/geronb/gbx143
27. Okamoto S. Socioeconomic factors and the risk of cognitive decline among the elderly population in Japan. International journal of geriatric psychiatry. 2019;34(2):265-71.Doi: https://10.1002/gps.5015
28. Al Hazzouri AZ, Haan MN, Kalbfleisch JD, et al. Life-course socioeconomic position and incidence of dementia and cognitive impairment without dementia in older Mexican Americans: Results from the Sacramento Area Latino Study on Aging. American Journal of Epidemiology. 2011;173(10):1148-58.Doi: https://10.1093/aje/kwq483
29. Alvarado BE, Zunzunegui MV, Del Ser T, Béland F. Cognitive decline is related to education and occupation in a Spanish elderly cohort. Aging Clin Exp Res. 2002;14(2):132-42.Doi: https://10.1007/bf03324426
30. Caldas V, Fernandes J, Vafaei A, et al. Life-Space and Cognitive Decline in Older Adults in Different Social and Economic Contexts: Longitudinal Results from the IMIAS Study. J Cross Cult Gerontol. 2020;35(3):237-54.Doi: https://10.1007/s10823-020-09406-8
31. Chen L, Cao Q. Poverty increases the risk of incident cognitive impairment among older adults: a longitudinal study in China. Aging Ment Health. 2020;24(11):1822-7.Doi: https://10.1080/13607863.2019.1663491
32. Chung W, Kim R. Differential Risk of Cognitive Impairment across Paid and Unpaid Occupations in the Middle-Age Population: Evidence from the Korean Longitudinal Study of Aging, 2006-2016. Int J Environ Res Public Health. 2020;17(9).Doi: https://10.3390/ijerph17093124
33. Kim Y. Risk and Protective Factors for the Onset of Cognitive Impairment in Korea: A 10-Year Longitudinal Panel Study. Psychiatry Investig. 2020;17(8):769-76.Doi: https://10.30773/pi.2020.0070
34. Lee S, Buring JE, Cook NR, Grodstein F. The relation of education and income to cognitive function among professional women. Neuroepidemiology. 2006;26(2):93-101.Doi: https://10.1159/000090254
35. Lee S, Kawachi I, Berkman LF, Grodstein F. Education, other socioeconomic indicators, and cognitive function. Am J Epidemiol. 2003;157(8):712-20.Doi: https://10.1093/aje/kwg042
36. Marengoni A, Fratiglioni L, Bandinelli S, Ferrucci L. Socioeconomic status during lifetime and cognitive impairment no-dementia in late life: the population-based aging in the Chianti Area (InCHIANTI) Study. J Alzheimers Dis. 2011;24(3):559-68.Doi: https://10.3233/jad-2011-101863
37. Sattler C, Toro P, Schönknecht P, Schröder J. Cognitive activity, education and socioeconomic status as preventive factors for mild cognitive impairment and Alzheimer’s disease. Psychiatry Res. 2012;196(1):90-5.Doi: https://10.1016/j.psychres.2011.11.012
38. Xu P, Burr JA, Kim K, Cong Z. Intergenerational financial exchange and cognitive impairment among older adults in China. Aging Ment Health. 2019;23(9):1209-17.Doi: https://10.1080/13607863.2018.1481930
39. Zhang Z, Gu D, Hayward MD. Early life influences on cognitive impairment among oldest old Chinese. J Gerontol B Psychol Sci Soc Sci. 2008;63(1):S25-33.Doi: https://10.1093/geronb/63.1.s25
40. Karp A, Kåreholt I, Qiu C, et al. Relation of education and occupation-based socioeconomic status to incident Alzheimer’s disease. Am J Epidemiol. 2004;159(2):175-83.Doi: https://10.1093/aje/kwh018
41. Koster A, Penninx BW, Bosma H, et al. Socioeconomic differences in cognitive decline and the role of biomedical factors. Ann Epidemiol. 2005;15(8):564-71.Doi: https://10.1016/j.annepidem.2005.02.008
42. Qiu C, Karp A, von Strauss E, et al. Lifetime principal occupation and risk of Alzheimer’s disease in the Kungsholmen project. Am J Ind Med. 2003;43(2):204-11.Doi: https://10.1002/ajim.10159
43. Cadar D, Lassale C, Davies H, et al. Individual and Area-Based Socioeconomic Factors Associated With Dementia Incidence in England: Evidence From a 12-Year Follow-up in the English Longitudinal Study of Ageing. JAMA Psychiatry. 2018;75(7):723-32.Doi: https://10.1001/jamapsychiatry.2018.1012
44. Deckers K, Cadar D, van Boxtel MPJ, et al. Modifiable Risk Factors Explain Socioeconomic Inequalities in Dementia Risk: Evidence from a Population-Based Prospective Cohort Study. J Alzheimers Dis. 2019;71(2):549-57.Doi: https://10.3233/jad-190541
45. Goldbourt U, Schnaider-Beeri M, Davidson M. Socioeconomic status in relationship to death of vascular disease and late-life dementia. J Neurol Sci. 2007;257(1-2):177-81.Doi: https://10.1016/j.jns.2007.01.021
46. Gureje O, Ogunniyi A, Kola L, Abiona T. Incidence of and risk factors for dementia in the Ibadan study of aging. J Am Geriatr Soc. 2011;59(5):869-74.Doi: https://10.1111/j.1532-5415.2011.03374.x
47. Helmer C, Letenneur L, Rouch I, et al. Occupation during life and risk of dementia in French elderly community residents. J Neurol Neurosurg Psychiatry. 2001;71(3):303-9.Doi: https://10.1136/jnnp.71.3.303
48. Kryscio RJ, Schmitt FA, Salazar JC, Mendiondo MS, Markesbery WR. Risk factors for transitions from normal to mild cognitive impairment and dementia. Neurology. 2006;66(6):828-32.Doi: https://10.1212/01.wnl.0000203264.71880.45
49. Kukull WA, Higdon R, Bowen JD, et al. Dementia and Alzheimer disease incidence: a prospective cohort study. Arch Neurol. 2002;59(11):1737-46.Doi: https://10.1001/archneur.59.11.1737
50. Lee JY, Chang SM, Jang HS, et al. Illiteracy and the incidence of Alzheimer’s disease in the Yonchon County survey, Korea. Int Psychogeriatr. 2008;20(5):976-85.Doi: https://10.1017/s1041610208007333
51. Myung W, Lee C, Park JH, et al. Occupational attainment as risk factor for progression from mild cognitive impairment to Alzheimer’s Disease: A CREDOS study. Journal of Alzheimer’s Disease. 2016;55(1):283-92.Doi: https://10.3233/JAD-160257
52. Ouvrard C, Meillon C, Dartigues JF, Ávila-Funes JA, Amieva H. Psychosocioeconomic Precariousness, Cognitive Decline and Risk of Developing Dementia: A 25-Year Study. Dement Geriatr Cogn Disord. 2016;41(3-4):137-45.Doi: https://10.1159/000443790
53. Park JE, Kim BS, Kim KW, et al. Decline in the Incidence of All-Cause and Alzheimer’s Disease Dementia: a 12-Year-Later Rural Cohort Study in Korea. J Korean Med Sci. 2019;34(44):e293.Doi: https://10.3346/jkms.2019.34.e293
54. Rusmaully J, Dugravot A, Moatti JP, et al. Contribution of cognitive performance and cognitive decline to associations between socioeconomic factors and dementia: A cohort study. PLoS Med. 2017;14(6):e1002334.Doi: https://10.1371/journal.pmed.1002334
55. Santabárbara J, Gracía-Rebled AC, López-Antón R, et al. The effect of occupation type on risk of Alzheimer’s disease in men and women. Maturitas. 2019;126:61-8.Doi: https://10.1016/j.maturitas.2019.05.008
56. Seblova D, Fischer M, Fors S, et al. Does Prolonged Education Causally Affect Dementia Risk When Adult Socioeconomic Status Is Not Altered? A Swedish Natural Experiment in 1.3 Million Individuals. Am J Epidemiol. 2021;190(5):817-26.Doi: https://10.1093/aje/kwaa255
57. Sindi S, Darin-Mattsson A, Kåreholt I, et al. Midlife improvements in financial situation are associated with a reduced dementia risk later in life: the CAIDE 30-year study. Int Psychogeriatr. 2020;32(11):1317-24.Doi: https://10.1017/s104161021900173x
58. Takasugi T, Tsuji T, Nagamine Y, Miyaguni Y, Kondo K. Socio-economic status and dementia onset among older Japanese: A 6-year prospective cohort study from the Japan Gerontological Evaluation Study. Int J Geriatr Psychiatry. 2019;34(11):1642-50.Doi: https://10.1002/gps.5177
59. Wilson RS, Scherr PA, Hoganson G, et al. Early life socioeconomic status and late life risk of Alzheimer’s disease. Neuroepidemiology. 2005;25(1):8-14.Doi: https://10.1159/000085307
60. Yaffe K, Falvey C, Harris TB, et al. Effect of socioeconomic disparities on incidence of dementia among biracial older adults: prospective study. Bmj. 2013;347:f7051.Doi: https://10.1136/bmj.f7051
61. Bickel H, Cooper B. Incidence and relative risk of dementia in an urban elderly population: findings of a prospective field study. Psychol Med. 1994;24(1):179-92.Doi: https://10.1017/s0033291700026945
62. Evans DA, Hebert LE, Beckett LA, et al. Education and other measures of socioeconomic status and risk of incident Alzheimer disease in a defined population of older persons. Arch Neurol. 1997;54(11):1399-405.Doi: https://10.1001/archneur.1997.00550230066019
63. Cheng HG, Shidhaye R, Charlson F, et al. Social correlates of mental, neurological, and substance use disorders in China and India: a review. Lancet Psychiatry. 2016;3(9):882-99.Doi: https://10.1016/s2215-0366(16)30166-3
64. Korous KM, Causadias JM, Bradley RH, Luthar SS, Levy R. A Systematic Overview of Meta-Analyses on Socioeconomic Status, Cognitive Ability, and Achievement: The Need to Focus on Specific Pathways. Psychol Rep. 2022;125(1):55-97.Doi: https://10.1177/0033294120984127
65. Lövdén M, Fratiglioni L, Glymour MM, Lindenberger U, Tucker-Drob EM. Education and Cognitive Functioning Across the Life Span. Psychol Sci Public Interest. 2020;21(1):6-41.Doi: https://10.1177/1529100620920576
66. Rosen J, Zubenko GS. Emergence of psychosis and depression in the longitudinal evaluation of Alzheimer’s disease. Biol Psychiatry. 1991;29(3):224-32.Doi: https://10.1016/0006-3223(91)91284-x
67. Stephan BCM, Pakpahan E, Siervo M, et al. Prediction of dementia risk in low-income and middle-income countries (the 10/66 Study): an independent external validation of existing models. Lancet Glob Health. 2020;8(4):e524-e35.Doi: https://10.1016/s2214-109x(20)30062-0
68. Alley D, Suthers K, Crimmins E. Education and Cognitive Decline in Older Americans: Results From the AHEAD Sample. Res Aging. 2007;29(1):73-94.Doi: https://10.1177/0164027506294245
69. Yang L, Martikainen P, Silventoinen K, Konttinen H. Association of socioeconomic status and cognitive functioning change among elderly Chinese people. Age Ageing. 2016;45(5):674-80.Doi: https://10.1093/ageing/afw107
70. Lynch JW, Kaplan GA, Shema SJ. Cumulative impact of sustained economic hardship on physical, cognitive, psychological, and social functioning. N Engl J Med. 1997;337(26):1889-95.Doi: https://10.1056/nejm199712253372606
71. Wilkinson RG. Health, hierarchy, and social anxiety. Ann N Y Acad Sci. 1999;896:48-63.Doi: https://10.1111/j.1749-6632.1999.tb08104.x
72. Albert MS. How does education affect cognitive function? Ann Epidemiol. 1995;5(1):76-8.Doi: https://10.1016/1047-2797(94)00044-t
73. Evans IEM, Martyr A, Collins R, Brayne C, Clare L. Social Isolation and Cognitive Function in Later Life: A Systematic Review and Meta-Analysis. J Alzheimers Dis. 2019;70(s1):S119-s44.Doi: https://10.3233/jad-180501
74. Bougea A, Anagnostouli M, Angelopoulou E, Spanou I, Chrousos G. Psychosocial and Trauma-Related Stress and Risk of Dementia: A Meta-Analytic Systematic Review of Longitudinal Studies. J Geriatr Psychiatry Neurol. 2022;35(1):24-37.Doi: https://10.1177/0891988720973759
75. Aisa B, Tordera R, Lasheras B, Del Río J, Ramírez MJ. Cognitive impairment associated to HPA axis hyperactivity after maternal separation in rats. Psychoneuroendocrinology. 2007;32(3):256-66.Doi: https://10.1016/j.psyneuen.2006.12.013
76. Katzman R. Education and the prevalence of dementia and Alzheimer’s disease. Neurology. 1993;43(1):13-20.Doi: https://10.1212/wnl.43.1_part_1.13
77. Gianaros PJ, Kuan DC, Marsland AL, et al. Community Socioeconomic Disadvantage in Midlife Relates to Cortical Morphology via Neuroendocrine and Cardiometabolic Pathways. Cereb Cortex. 2017;27(1):460-73.Doi: https://10.1093/cercor/bhv233
78. Farah MJ. The Neuroscience of Socioeconomic Status: Correlates, Causes, and Consequences. Neuron. 2017;96(1):56-71.Doi: https://10.1016/j.neuron.2017.08.034
79. Holz NE, Boecker R, Hohm E, et al. The long-term impact of early life poverty on orbitofrontal cortex volume in adulthood: results from a prospective study over 25 years. Neuropsychopharmacology. 2015;40(4):996-1004.Doi: https://10.1038/npp.2014.277
80. Farah MJ. Socioeconomic status and the brain: prospects for neuroscience-informed policy. Nat Rev Neurosci. 2018;19(7):428-38.Doi: https://10.1038/s41583-018-0023-2
81. Muurinen JM. Demand for health: a generalised Grossman model. J Health Econ. 1982;1(1):5-28.Doi: https://10.1016/0167-6296(82)90019-4
82. Hartwig J, Sturm JE. Testing the Grossman model of medical spending determinants with macroeconomic panel data. Eur J Health Econ. 2018;19(8):1067-86.Doi: https://10.1007/s10198-018-0958-2
83. Adler NE, Newman K. Socioeconomic disparities in health: pathways and policies. Health Aff (Millwood). 2002;21(2):60-76.Doi: https://10.1377/hlthaff.21.2.60