Abstract
Introduction: The prevalence of mild and major neurocognitive disorders (NCDs), also referred to as mild cognitive impairment and dementia, is rising globally. The prevention of NCDs is a major global public health interest. We sought to synthesize the literature on potentially modifiable risk factors for NCDs. Methods: We conducted an umbrella review using a systematic search across multiple databases to identify relevant systematic reviews and meta-analyses. Eligible reviews examined potentially modifiable risk factors for mild or major NCDs. We used a random-effects multi-level meta-analytic approach to synthesize risk ratios for each risk factor while accounting for overlap in the reviews. We further examined risk factors for major NCD due to two common etiologies: Alzheimer’s disease and vascular dementia. Results: A total of 45 reviews with 212 meta-analyses were synthesized. We identified fourteen broadly defined modifiable risk factors that were significantly associated with these disorders: alcohol consumption, body weight, depression, diabetes mellitus, diet, hypertension, less education, physical inactivity, sensory loss, sleep disturbance, smoking, social isolation, traumatic brain injury, and vitamin D deficiency. All 14 factors were associated with the risk of major NCD, and five were associated with mild NCD. We found considerably less research for vascular dementia and mild NCD. Conclusion: Our review quantifies the risk associated with 14 potentially modifiable risk factors for mild and major NCDs, including several factors infrequently included in dementia action plans. Prevention strategies should consider approaches that reduce the incidence and severity of these risk factors through health promotion, identification, and early management.
Introduction
Neurocognitive disorders (NCDs) are a leading cause of mortality and morbidity worldwide and a major subject of biomedical, clinical, and public health research [1]. NCDs were included in the Fifth Edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), which distinguished between mild and major forms [2]. It was hoped that the term “major NCD” would be viewed as less stigmatizing than “dementia” while the inclusion of criteria for a milder form of acquired cognitive decline would provide opportunities for earlier detection, advance care planning, and treatment before the condition progresses to a severe stage [3]. Major NCDs are marked by significant cognitive decline that results in impairments of activities of daily living, which are not better explained by delirium or another mental health disorder. Mild NCDs, also referred to as mild cognitive impairment in the neurology literature, are characterized by more modest cognitive decline that does not result in functional impairment with the same exclusionary criteria noted for major NCDs. While mild NCD is not necessarily progressive, between 30% and 50% of older adults with mild degrees of impaired cognition progress to a major NCD within 5 years [4‒6]. Both major and mild NCDs can be secondary to a variety of conditions including Alzheimer’s disease (AD) and vascular cognitive impairment (VCI).
Major NCD was estimated to affect 57.4 million individuals worldwide in 2019, a figure that is projected to increase to 152.8 million by 2050 (a 266% increase) [7]. This increase is due to the global growth of the older population offsetting declines in age-specific incidence rates in economically developed nations. While estimates of the prevalence of mild NCD vary widely from factors such as the population studied and definition used, a recent umbrella review found that, globally, approximately 15% of adults over the age of 50 were living with this condition [8]. The typical progressive nature of major NCDs, stage-dependent mortality and morbidity risk, and the limited treatment options has led governmental and public health organizations to emphasize the importance of its prevention [9, 10]. There is an extensive scientific literature on risk factors for major NCD that has been summarized in the Lancet Commission on Dementia Prevention, Intervention, and Care reports [11, 12]. Many of the risk factors identified for dementia are potentially modifiable, but whether this is also true for mild NCDs is less clear as the evidence on risk factors for this condition is less voluminous. Few reports to date have simultaneously examined the literature for risk factors of both major and mild NCDs.
Given that mild and major NCDs represent a continuum of severity with similar etiologies, many of their risk factors are likely shared. Evidence related to the overlap in risk factors, though, remains limited. As well, shared risk factors could still vary meaningfully in their relative importance to the development and/or progression of each condition. A synthesis of the modifiable risk factors for mild and major NCDs could inform prevention strategies that encompass the continuum. The objective of this systematic review of systematic reviews (i.e., umbrella review) and meta-analysis is to examine and synthesize the existing literature on modifiable risk factors for mild and major NCDs. We will compare the evidence on risk factors for both mild and major NCDs and for two of the more common etiologies of major NCDs.
Methods
This umbrella review and meta-analysis is reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the JBI Manual for Evidence Synthesis [13, 14]. The protocol for this review was registered a priori in PROSPERO (CRD42021291607). We chose an umbrella review methodology given the large number of existing systematic reviews found in the literature.
Eligibility Criteria
Publications eligible for inclusion were systematic reviews with meta-analyses that examined risk factors for mild or major NCDs. No specific diagnostic criteria were enforced and we did not exclude any etiologies. Publications were excluded if they were in languages other than English, did not include longitudinal data, or primarily considered populations living in congregate living settings (e.g., nursing homes, etc.). Studies were ineligible if their outcome was cognitive decline apart from a new diagnosis of an NCD. Due to the large number of risk factors uncovered by the review, the present study considers only potentially modifiable risk factors. Non-modifiable risk factors, such as age, sex/gender, and family history, were not included. Potentially modifiable environmental and occupational (e.g., air pollution, pesticides, etc.) risk factors and early life adversity were also not considered in this review.
Study Selection
We searched MEDLINE (Ovid MEDLINE [R]), Web of Science (core collection), PsycINFO (Ovid interface), and Cochrane Library (Wiley online platform) for relevant articles from January 1984 until November 2021. An electronic search strategy (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000536643) was developed in consultation with academic librarians. The search string included terms related to systematic reviews and/or meta-analyses, identification of risk factors, and NCDs. After removal of duplicates, studies were first screened for relevance by title and abstract and then were assessed for eligibility by full-text review. Screening and eligibility assessment was performed by two independent reviewers; conflicts were resolved by a third reviewer. Risk factors were sorted by modifiable/non-modifiable after screening and eligibility, with only the potentially modifiable risk factors retained for this review. The scope of potentially modifiable risk factors was determined through consideration of previous literature [15], public health guidelines [16], and consensus within the research team.
Data Extraction and Quality Assessment
Quality assessment and data extraction were performed by a single reviewer and verified by a second reviewer. Quality was assessed using the AMSTAR-2 critical appraisal tool [17]. The AMSTAR-2 tool rates the evidence from a systematic review on 16 items across 10 domains. The tool is not intended to generate an overall score but rather a rating from very low to high based on the number of critical and non-critical weaknesses. Data extracted included year of publication, number of studies included, risk factors examined, outcomes examined, and number of participants. Study results extracted included the name and definition of each risk factor, name and definition of each outcome, and the meta-analyzed risk estimates with associated standard error, and/or 95% confidence interval. Outcomes were classified into major NCD (dementia) and mild NCD (mild cognitive impairment). Results for subgroups, including for dementia due to AD and dementia due to VCI (i.e., vascular dementia), were extracted where available.
Data Synthesis
We reported the characteristics of the included studies using descriptive statistics. For consistency in synthesizing results across reviews, we used relative risk (RR) to quantify the association between modifiable risk factors and the risk of NCDs. Hazard ratios were used as equivalent to risk ratios and odds ratios were converted to risk ratios from longitudinal/prospective studies using Cochrane recommended methods [18, 19]. For meta-analysis, we used a random-effects multi-level meta-analytic approach to account for dependency between effect estimates across reviews and to avoid possible unit of analysis error (i.e., a correlation between effect estimates due to overlap of studies across reviews or multiple measures/comparisons of the outcome within the same review with overlapping sub-sample sizes or population) [20, 21]. In such cases, possibly correlated effect estimates were nested at the within-review level first (reviews with overlapping studies or multiple measures from the same review). The variance in observed effect estimates was further decomposed into sampling variance, within-review variance, and between-review variance to account for intracluster (or intraclass) correlation in the true effects. The final pooled effect estimates were presented with 95% confidence intervals, which were used to judge statistical significance. When there was only a single systematic review available for a given association, we did not perform additional meta-analysis but reported the result of the underlying systematic review.
Overall heterogeneity for each summary effect estimate was estimated using the I2 statistic, which represents the heterogeneity not attributable to sample error and is the sum of within-cluster (i.e., across effect sizes or multiple arms from the same or overlapping reviews) and between-cluster heterogeneity (i.e., effect sizes across unique reviews) [22]. We considered I2 values of between 50% and 75% to represent moderate heterogeneity across studies and I2 values of >75% to represent substantial heterogeneity across studies. Cochran’s Q (α = 0.05) was also employed to detect statistical heterogeneity.
The primary grouping in the meta-analysis was based on risk factor, then by the type of NCD and further by the two of the more common causes of major NCD. To account for expected heterogeneity across life stages, analysis of body weight and hypertension was conducted by age-groups (mixed age-groups, mid-life [40–64], and late life [65+]). For body weight, body mass index (BMI) (kg/m2) categories were as follows: <18.5, ≥18.5 and <25, ≥25 and <30, and ≥30. The reference category was between 18.5 and 25 kg/m2. There were insufficient reported data across reviews to perform additional subgroup analyses (i.e., based on sex/gender, socio-economic status, length of follow-up, etc.). All analyses were performed using R v4.3.1 with packages metafor and dmetar packages [23, 24].
Results
Literature Flow
After removing duplicates, 6,790 articles were screened via title and abstract, and 1,114 were assessed for eligibility at the full-text level (Fig. 1). Ultimately 45 [25‒69] articles were included with containing 212 meta-analyses with 89,671,556 participants examining 14 broadly defined risk factors.
Study Characteristics
Eligible articles were published between 2006 and 2021 (Table 1). Of the 212 meta-analyses included, 86 examined the risk of major NCD, 83 examined dementia from AD, 29 examined dementia from VCI (i.e., vascular dementia), and 14 examined mild NCD. The broadly defined risk factors included alcohol consumption, body weight (BMI <18.5, BMI 25–29, BMI ≥30), depression, diabetes mellitus, diet (fish consumption, milk consumption, healthy dietary patterns), hypertension (combined, systolic, diastolic), less education, physical inactivity, sensory loss (hearing loss, visual impairment), sleep disturbances, smoking, social isolation, traumatic brain injury (TBI), vitamin D deficiency. Risk factor definitions varied within and between reviews (online suppl. Table S2) with the exception of body weight, which used standard BMI categories, and smoking, which consistently used current, former, and ever smoker categories. Health conditions were most commonly based on validated cut-offs of objective measures and/or scales but also included diagnoses noted in clinical records and self-report. Social isolation was primarily defined as living alone. Sleep disturbances included a variety of sleeping disorders as well as self-reported long and short sleep durations. Visual impairment included age-related macular degeneration as well as objective loss of visual acuity. Less education did not have a consistent definition but was related to fewer years of formal schooling.
Characteristics of included reviews (n = 45)
Review . | Risk factors examined . | Outcomes examined . | Number of studies included . | Number of participants included across studies . |
---|---|---|---|---|
Anstey [25] (2007) | Smoking | Major NCD; Alzheimer’s dementia, vascular dementia | 8 | 22,441 |
Anstey [26] (2009) | Alcohol consumption | Major NCD; Alzheimer’s dementia, vascular dementia | 15 | 29,686 |
Anstey [27] (2011) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 17 | 46,462 |
Bakre [28] (2018) | Diet | Major NCD; Alzheimer’s dementia | 11 | 33,964 |
Beydoun [29] (2008) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 10 | 41,956 |
Bubu [30] (2017) | Sleep disturbance | Alzheimer’s dementia | 27 | 69,216 |
Cherbuin [31] (2015) | Depression | Major NCD; Alzheimer’s dementia, vascular dementia | 36 | 66,532 |
Cooper [32] (2015) | Diabetes; hypertension; depression | Major NCD | 62 | 15,223 |
Danat [33] (2019) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 16 | 41,781 |
Desai [34] (2020) | Social isolation | Major NCD | 12 | 13,317 |
Hamer [35] (2009) | Physical inactivity | Major NCD; Alzheimer’s dementia | 16 | 163,797 |
Hu [36] (2020) | Depression | Mild NCD | 13 | 33,066 |
Huang [37] (2013) | TBI | Major NCD; Alzheimer’s dementia | 18 | 4,648,013 |
Hudon [38] (2020) | Sleep disturbance, depression | Major NCD; mild NCD | 18 | 72,287 |
Kalra [39] (2020) | Vitamin D deficiency | Major NCD; Alzheimer’s dementia | 5 | 19,259 |
Kuring [40] (2020) | Depression | Major NCD; Alzheimer’s dementia, vascular dementia | 36 | 828,767 |
Lee [41] (2020) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 23 | 2,790,753 |
Lennon [42] (2019) | Hypertension | Alzheimer’s dementia | 7 | 877,321 |
Li [43] (2019) | Diabetes; hypertension; body weight; smoking | Major NCD | 24 | 159,594 |
Liang [44] (2020) | Alcohol consumption; sleep disturbance; diabetes; low education; hypertension; social isolation; body weight; physical inactivity; smoking | Major NCD | 43 | 277,294 |
Liang [45] (2021) | Sensory loss | Major NCD; Alzheimer’s dementia | 14 | 726,900 |
Loef [46] (2013) | Body weight | Major NCD; Alzheimer’s dementia | 13 | 59,978 |
Loughrey [47] (2018) | Sensory loss | Major NCD; Alzheimer’s dementia, vascular dementia | 40 | 34,471 |
Mourao [48] (2016) | Depression | Major NCD | 18 | 10,861 |
Ou [49] (2020) | Hypertension | Major NCD; Alzheimer’s dementia, mild NCD | 136 | 2,214,814 |
Ownby [50] (2006) | Depression | Alzheimer’s dementia | 20 | 102,172 |
Pedditizi [51] (2016-A) | Body weight | Major NCD | 21 | 62,425 |
Perry [52] (2016) | TBI | Major NCD; mild NCD | 30 | 8,232 |
Power [53] (2011) | Hypertension | Alzheimer’s dementia | 18 | 42,986 |
Profenno [54] (2010) | Diabetes; body weight | Alzheimer’s dementia | 15 | 41,480 |
Qu [55] (2020) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 18 | 201,026 |
Rong [56] (2019) | Sensory loss | Alzheimer’s dementia | 21 | 7,876,499 |
Sáiz-Vázquez [57] (2021) | Depression | Alzheimer’s dementia | 28 | 50,051 |
Snowden [58] (2020) | TBI | Major NCD | 5 | 2,351,334 |
Sommerlad [59] (2018) | Social isolation | Major NCD | 15 | 812,047 |
Tan [60] (2019) | Depression | Major NCD | 35 | 14,158 |
Vagelatos [61] (2013) | Diabetes | Alzheimer’s dementia | 15 | 2,122,883 |
Vu [62] (2021) | Sensory loss | Major NCD | 40 | 47,913,570 |
Wang [63] (2018) | Hypertension | Major NCD | 21 | 830,631 |
Wei [64] (2017) | Sensory Loss | Mild NCD | 10 | 15,521 |
Wu [65] (2016) | Diet | Major NCD; Alzheimer’s dementia | 7 | 10,941 |
Xu [66] (2015) | Body weight, smoking, alcohol consumption, hypertension, diabetes, depression, low education, diet, physical inactivity, social isolation, TBI | Alzheimer’s dementia | 351 | 3,116,095 |
Yates [67] (2013) | Depression | Major NCD; MCI | 60 | 53,451 |
Zhong [68] (2015) | Smoking | Major NCD; Alzheimer’s dementia, vascular dementia | 37 | 960,280 |
Zuin [69] (2021) | Diabetes; hypertension | Alzheimer’s dementia | 6 | 9,788,021 |
Review . | Risk factors examined . | Outcomes examined . | Number of studies included . | Number of participants included across studies . |
---|---|---|---|---|
Anstey [25] (2007) | Smoking | Major NCD; Alzheimer’s dementia, vascular dementia | 8 | 22,441 |
Anstey [26] (2009) | Alcohol consumption | Major NCD; Alzheimer’s dementia, vascular dementia | 15 | 29,686 |
Anstey [27] (2011) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 17 | 46,462 |
Bakre [28] (2018) | Diet | Major NCD; Alzheimer’s dementia | 11 | 33,964 |
Beydoun [29] (2008) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 10 | 41,956 |
Bubu [30] (2017) | Sleep disturbance | Alzheimer’s dementia | 27 | 69,216 |
Cherbuin [31] (2015) | Depression | Major NCD; Alzheimer’s dementia, vascular dementia | 36 | 66,532 |
Cooper [32] (2015) | Diabetes; hypertension; depression | Major NCD | 62 | 15,223 |
Danat [33] (2019) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 16 | 41,781 |
Desai [34] (2020) | Social isolation | Major NCD | 12 | 13,317 |
Hamer [35] (2009) | Physical inactivity | Major NCD; Alzheimer’s dementia | 16 | 163,797 |
Hu [36] (2020) | Depression | Mild NCD | 13 | 33,066 |
Huang [37] (2013) | TBI | Major NCD; Alzheimer’s dementia | 18 | 4,648,013 |
Hudon [38] (2020) | Sleep disturbance, depression | Major NCD; mild NCD | 18 | 72,287 |
Kalra [39] (2020) | Vitamin D deficiency | Major NCD; Alzheimer’s dementia | 5 | 19,259 |
Kuring [40] (2020) | Depression | Major NCD; Alzheimer’s dementia, vascular dementia | 36 | 828,767 |
Lee [41] (2020) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 23 | 2,790,753 |
Lennon [42] (2019) | Hypertension | Alzheimer’s dementia | 7 | 877,321 |
Li [43] (2019) | Diabetes; hypertension; body weight; smoking | Major NCD | 24 | 159,594 |
Liang [44] (2020) | Alcohol consumption; sleep disturbance; diabetes; low education; hypertension; social isolation; body weight; physical inactivity; smoking | Major NCD | 43 | 277,294 |
Liang [45] (2021) | Sensory loss | Major NCD; Alzheimer’s dementia | 14 | 726,900 |
Loef [46] (2013) | Body weight | Major NCD; Alzheimer’s dementia | 13 | 59,978 |
Loughrey [47] (2018) | Sensory loss | Major NCD; Alzheimer’s dementia, vascular dementia | 40 | 34,471 |
Mourao [48] (2016) | Depression | Major NCD | 18 | 10,861 |
Ou [49] (2020) | Hypertension | Major NCD; Alzheimer’s dementia, mild NCD | 136 | 2,214,814 |
Ownby [50] (2006) | Depression | Alzheimer’s dementia | 20 | 102,172 |
Pedditizi [51] (2016-A) | Body weight | Major NCD | 21 | 62,425 |
Perry [52] (2016) | TBI | Major NCD; mild NCD | 30 | 8,232 |
Power [53] (2011) | Hypertension | Alzheimer’s dementia | 18 | 42,986 |
Profenno [54] (2010) | Diabetes; body weight | Alzheimer’s dementia | 15 | 41,480 |
Qu [55] (2020) | Body weight | Major NCD; Alzheimer’s dementia, vascular dementia | 18 | 201,026 |
Rong [56] (2019) | Sensory loss | Alzheimer’s dementia | 21 | 7,876,499 |
Sáiz-Vázquez [57] (2021) | Depression | Alzheimer’s dementia | 28 | 50,051 |
Snowden [58] (2020) | TBI | Major NCD | 5 | 2,351,334 |
Sommerlad [59] (2018) | Social isolation | Major NCD | 15 | 812,047 |
Tan [60] (2019) | Depression | Major NCD | 35 | 14,158 |
Vagelatos [61] (2013) | Diabetes | Alzheimer’s dementia | 15 | 2,122,883 |
Vu [62] (2021) | Sensory loss | Major NCD | 40 | 47,913,570 |
Wang [63] (2018) | Hypertension | Major NCD | 21 | 830,631 |
Wei [64] (2017) | Sensory Loss | Mild NCD | 10 | 15,521 |
Wu [65] (2016) | Diet | Major NCD; Alzheimer’s dementia | 7 | 10,941 |
Xu [66] (2015) | Body weight, smoking, alcohol consumption, hypertension, diabetes, depression, low education, diet, physical inactivity, social isolation, TBI | Alzheimer’s dementia | 351 | 3,116,095 |
Yates [67] (2013) | Depression | Major NCD; MCI | 60 | 53,451 |
Zhong [68] (2015) | Smoking | Major NCD; Alzheimer’s dementia, vascular dementia | 37 | 960,280 |
Zuin [69] (2021) | Diabetes; hypertension | Alzheimer’s dementia | 6 | 9,788,021 |
The median number of studies per review was 18 (13–28), and the median number of participants per review across all reviews was 65,532 (33,066–830,631). Our quality appraisal of the literature using AMSTAR-2 found that 18 (40%) reviews were of high quality, 10 (22%) were of moderate quality, 8 (18%) were of low quality, and 9 (20%) were of very low quality (online suppl. Table S3).
Risk Factors for Major NCD
Major NCD was the most common outcome examined. We extracted results for 14 risk factors (Table 2) from 86 meta-analyses. There were statistically significant positive associations between major NCD and body weight (BMI ≥30, BMI 25–29, and BMI <18.5 at mid-life; BMI <18.5 in a mixed age-group), depression, diabetes mellitus, diet (low fish consumption), hypertension (combined hypertension at mid-life and in mixed age-groups; high systolic blood pressure at mid-life and in mixed age-groups; high diastolic blood pressure in mid-life), less education, physical inactivity, sensory loss (hearing loss, visual impairment), sleep disturbances, ever or current smoking, social isolation, TBI, vitamin D deficiency. The largest risk ratio was observed with sleep disturbances at 2.18 (95% CI, 1.44, 3.32). Heavy alcohol consumption was not a significant risk factor compared to no alcohol consumption but was a significant risk factor when compared to light/moderate alcohol consumption. Factors that were protective included BMI ≥30 and BMI 25–29 in late life, BMI 25–29 in the mixed age-group, and any or light/moderate alcohol intake versus no alcohol intake. Of the 25 associations for which the I2 statistic could be calculated, 1 study (4%) had a value between 50% and 75% and 4 (16%) had I2 ≥ 75%.
Results of meta-analysis of risk factors for major NCD
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Alcohol consumption | |||
Any versus none | 1 | 0.66 (0.53, 0.82) | - |
Light/moderate versus none | 1 | 0.74 (0.61, 0.90) | - |
Heavy versus none | 2 | 1.19 (0.95, 1.50) | 0% |
Heavy versus light/moderate | 1 | 1.41 (1.14, 1.73) | - |
Body weight (mid-life) | |||
BMI ≥30 versus normal | 6 | 1.57 (1.43, 1.72) | 4% |
BMI 25–29.9 versus normal | 5 | 1.19 (1.12, 1.27) | 4% |
BMI <18.5 versus normal | 1 | 1.42 (1.12, 1.80) | - |
Body weight (late life) | |||
BMI ≥30 versus normal | 3 | 0.80 (0.74, 0.87) | 0% |
BMI 25–29.9 versus normal | 3 | 0.84 (0.89, 0.92) | 0% |
BMI <18.5 versus normal | 2 | 1.11 (0.85, 1.46) | 49% |
Body weight (mixed age-groups) | |||
BMI ≥30 versus normal | 3 | 1.23 (0.82, 1.85) | 84% |
BMI 25–29.9 versus normal | 2 | 0.86 (0.81, 0.92) | 0% |
BMI <18.5 versus normal | 2 | 1.26 (1.21, 1.32) | 0% |
Depression | 6 | 1.41 (1.16, 1.72) | 87% |
Diabetes mellitus | 3 | 1.67 (1.43, 1.95) | 0% |
Diet | |||
Low fish consumption versus high | 1 | 1.25 (1.15, 1.35) | - |
Low milk consumption versus high | 1 | 1.37 (0.98, 1.96) | - |
Hypertension (mid-life) | |||
Hypertension (combined) | 3 | 1.51 (1.11, 2.04) | 77% |
High systolic pressure | 3 | 1.51 (1.31, 1.74) | 0% |
High diastolic pressure | 2 | 1.54 (1.16, 2.05) | 0% |
Hypertension (late life) | |||
Hypertension (combined) | 2 | 1.03 (0.95, 1.11) | 0% |
High systolic pressure | 2 | 1.09 (0.91, 1.30) | 0% |
High diastolic pressure | 2 | 0.86 (0.64, 1.15) | 32% |
Hypertension (mixed age-groups) | |||
Hypertension (combined) | 2 | 1.20 (1.04, 1.40) | 0% |
High systolic pressure | 1 | 1.20 (1.02, 1.42) | - |
High diastolic pressure | 1 | 1.24 (0.85, 1.81) | - |
Less education | 1 | 1.87 (1.40, 2.48) | - |
Physical inactivity | 2 | 1.37 (1.17, 1.60) | 0% |
Sensory loss | |||
Hearing loss | 3 | 1.59 (1.20, 2.11) | 78% |
Visual impairment | 1 | 2.05 (1.36, 3.09) | - |
Sleep disturbance | 1 | 2.18 (1.44, 3.32) | - |
Smoking | |||
Ever smokers | 3 | 1.22 (1.04, 1.43) | 57% |
Current smokers | 3 | 1.37 (1.20, 1.57) | 49% |
Former smokers | 3 | 1.01 (0.97, 1.04) | 0% |
Social isolation | 5 | 1.22 (1.15, 1.29) | 23% |
TBI | 2 | 1.92 (1.70, 2.17) | 0% |
Vitamin D deficiency | 1 | 1.33 (1.15, 1.54) | - |
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Alcohol consumption | |||
Any versus none | 1 | 0.66 (0.53, 0.82) | - |
Light/moderate versus none | 1 | 0.74 (0.61, 0.90) | - |
Heavy versus none | 2 | 1.19 (0.95, 1.50) | 0% |
Heavy versus light/moderate | 1 | 1.41 (1.14, 1.73) | - |
Body weight (mid-life) | |||
BMI ≥30 versus normal | 6 | 1.57 (1.43, 1.72) | 4% |
BMI 25–29.9 versus normal | 5 | 1.19 (1.12, 1.27) | 4% |
BMI <18.5 versus normal | 1 | 1.42 (1.12, 1.80) | - |
Body weight (late life) | |||
BMI ≥30 versus normal | 3 | 0.80 (0.74, 0.87) | 0% |
BMI 25–29.9 versus normal | 3 | 0.84 (0.89, 0.92) | 0% |
BMI <18.5 versus normal | 2 | 1.11 (0.85, 1.46) | 49% |
Body weight (mixed age-groups) | |||
BMI ≥30 versus normal | 3 | 1.23 (0.82, 1.85) | 84% |
BMI 25–29.9 versus normal | 2 | 0.86 (0.81, 0.92) | 0% |
BMI <18.5 versus normal | 2 | 1.26 (1.21, 1.32) | 0% |
Depression | 6 | 1.41 (1.16, 1.72) | 87% |
Diabetes mellitus | 3 | 1.67 (1.43, 1.95) | 0% |
Diet | |||
Low fish consumption versus high | 1 | 1.25 (1.15, 1.35) | - |
Low milk consumption versus high | 1 | 1.37 (0.98, 1.96) | - |
Hypertension (mid-life) | |||
Hypertension (combined) | 3 | 1.51 (1.11, 2.04) | 77% |
High systolic pressure | 3 | 1.51 (1.31, 1.74) | 0% |
High diastolic pressure | 2 | 1.54 (1.16, 2.05) | 0% |
Hypertension (late life) | |||
Hypertension (combined) | 2 | 1.03 (0.95, 1.11) | 0% |
High systolic pressure | 2 | 1.09 (0.91, 1.30) | 0% |
High diastolic pressure | 2 | 0.86 (0.64, 1.15) | 32% |
Hypertension (mixed age-groups) | |||
Hypertension (combined) | 2 | 1.20 (1.04, 1.40) | 0% |
High systolic pressure | 1 | 1.20 (1.02, 1.42) | - |
High diastolic pressure | 1 | 1.24 (0.85, 1.81) | - |
Less education | 1 | 1.87 (1.40, 2.48) | - |
Physical inactivity | 2 | 1.37 (1.17, 1.60) | 0% |
Sensory loss | |||
Hearing loss | 3 | 1.59 (1.20, 2.11) | 78% |
Visual impairment | 1 | 2.05 (1.36, 3.09) | - |
Sleep disturbance | 1 | 2.18 (1.44, 3.32) | - |
Smoking | |||
Ever smokers | 3 | 1.22 (1.04, 1.43) | 57% |
Current smokers | 3 | 1.37 (1.20, 1.57) | 49% |
Former smokers | 3 | 1.01 (0.97, 1.04) | 0% |
Social isolation | 5 | 1.22 (1.15, 1.29) | 23% |
TBI | 2 | 1.92 (1.70, 2.17) | 0% |
Vitamin D deficiency | 1 | 1.33 (1.15, 1.54) | - |
Risk Factors for Mild NCD
We extracted results from 14 meta-analyses concerning mild NCD for 5 risk factors (Table 3). We found statistically significant positive associations between risk of mild NCD and depression, hearing loss, hypertension (combined hypertension at mid- and late life, high systolic pressure at mid-life, high diastolic pressure at mid-life), sleep disturbances, and TBI. The highest risk ratio was observed with TBI at 2.51 (1.50, 4.18). No protective factors were found. The I2 statistic could be calculated for 2 associations, one of which (50%) was greater than 50% but less than 75%.
Results of meta-analysis of risk factors for mild NCD
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Depression | 4 | 1.80 (1.40, 2.31) | 69% |
Hypertension (mid-life) | |||
Hypertension (combined) | 1 | 1.55 (1.19, 2.02) | - |
High systolic pressure | 1 | 2.17 (1.44, 3.27) | - |
High diastolic pressure | 1 | 1.09 (1.02, 1.17) | - |
Hypertension (late life) | |||
Hypertension (combined) | 1 | 1.26 (1.04, 1.53) | - |
High systolic pressure | 1 | 1.34 (0.93, 1.93) | - |
High diastolic pressure | 1 | 0.97 (0.89, 1.06) | - |
Sensory loss | |||
Hearing loss | 1 | 1.30 (1.12, 1.51) | - |
Sleep disturbance | 2 | 1.27 (1.12, 1.43) | 0% |
TBI | 1 | 2.51 (1.50, 4.18) | - |
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Depression | 4 | 1.80 (1.40, 2.31) | 69% |
Hypertension (mid-life) | |||
Hypertension (combined) | 1 | 1.55 (1.19, 2.02) | - |
High systolic pressure | 1 | 2.17 (1.44, 3.27) | - |
High diastolic pressure | 1 | 1.09 (1.02, 1.17) | - |
Hypertension (late life) | |||
Hypertension (combined) | 1 | 1.26 (1.04, 1.53) | - |
High systolic pressure | 1 | 1.34 (0.93, 1.93) | - |
High diastolic pressure | 1 | 0.97 (0.89, 1.06) | - |
Sensory loss | |||
Hearing loss | 1 | 1.30 (1.12, 1.51) | - |
Sleep disturbance | 2 | 1.27 (1.12, 1.43) | 0% |
TBI | 1 | 2.51 (1.50, 4.18) | - |
Risk Factors for AD Dementia
We extracted results from the 83 meta-analyses related to AD dementia for 14 risk factors (Table 4). We found statistically significant positive associations between risk of AD dementia and body weight (BMI ≥30 and BMI 25–29 at mid-life; BMI <18.5 in a mixed age-group), depression, diabetes mellitus, diet (low fish consumption, low milk consumption), hypertension (combined hypertension at mid-life and in mixed age-groups; high systolic pressure in mixed age-group; high diastolic pressure at mid-life), less education, physical inactivity, sensory loss (hearing loss), sleep disturbances, ever or current smoking, and vitamin D deficiency. The largest risk ratio was observed with BMI ≥30 at mid-life at 2.13 (1.75, 2.59). Heavy alcohol consumption was not a significant risk factor compared to no alcohol consumption but was a significant risk factor when compared to light/moderate alcohol consumption. Factors that were found to be protective included BMI ≥30 and BMI 25–29 at late life and BMI 25–29 in mixed age-groups, any and light/moderate alcohol intake versus no alcohol intake, and healthy dietary patterns. Of the 26 associations for which the I2 statistics could be calculated, 3 (12%) had I2 between 50% and 75% and 4 (15%) had I2 ≥ 75%.
Results of meta-analysis of risk factors for AD dementia
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Alcohol consumption | |||
Any versus none | 2 | 0.60 (0.42, 0.85) | 13% |
Light/moderate versus none | 2 | 0.65 (0.56, 0.77) | 60% |
Heavy versus none | 1 | 1.36 (0.68, 2.72) | - |
Heavy versus light/moderate | 1 | 1.89 (1.12, 3.18) | - |
Body weight (mid-life) | |||
BMI ≥30 versus normal | 5 | 2.13 (1.75, 2.59) | 52% |
BMI 25–29.9 versus normal | 3 | 1.40 (1.25, 1.57) | 0% |
BMI <18.5 versus normal | 3 | 1.43 (0.95, 2.17) | 57% |
Body weight (late life) | |||
BMI ≥30 versus normal | 3 | 0.75 (0.67, 0.93) | 4% |
BMI 25–29.9 versus normal | 2 | 0.71 (0.59, 0.84) | 0% |
BMI <18.5 versus normal | 1 | 1.17 (0.70, 1.94) | - |
Body weight (mixed age-groups) | |||
BMI ≥30 versus normal | 3 | 1.30 (0.80, 2.11) | 79% |
BMI 25–29.9 versus normal | 1 | 0.89 (0.81, 0.98) | - |
BMI <18.5 versus normal | 2 | 1.37 (1.20, 1.57) | 6% |
Depression | 6 | 1.64 (1.20, 2.23) | 92% |
Diabetes mellitus | 4 | 1.39 (1.18, 1.65) | 92% |
Diet | |||
Low fish consumption versus high | 2 | 1.39 (1.23, 1.54) | 0% |
Low milk consumption versus high | 1 | 1.54 (1.10, 2.17) | - |
Healthy dietary patterns | 1 | 0.46 (0.27, 0.79) | - |
Hypertension (mid-life) | |||
Hypertension (combined) | 3 | 1.19 (1.08, 1.32) | 0% |
High systolic pressure | 2 | 1.37 (0.97, 1.92) | 0% |
High diastolic pressure | 2 | 1.53 (1.12, 2.10) | 0% |
Hypertension (late life) | |||
Hypertension (combined) | 3 | 0.91 (0.82, 1.02) | 0% |
High systolic pressure | 2 | 0.97 (0.81, 1.16) | 0% |
High diastolic pressure | 2 | 0.79 (0.54, 1.14) | 0% |
Hypertension (mixed age-groups) | |||
Hypertension | 2 | 1.05 (1.04, 1.06) | 0% |
High systolic pressure | 3 | 1.04 (1.01, 1.06) | 75% |
High diastolic pressure | 1 | 1.33 (0.62, 2.86) | - |
Less education | 1 | 1.35 (1.14, 1.60) | - |
Physical inactivity | 2 | 1.56 (1.28, 1.90) | 0% |
Sensory loss | |||
Hearing loss | 2 | 2.03 (1.31, 3.14) | 0% |
Visual impairment | 1 | 1.27 (0.53, 3.04) | - |
Sleep disturbance | 1 | 1.55 (1.25, 1.93) | - |
Smoking | |||
Ever smokers | 2 | 1.12 (1.00, 1.26) | 0% |
Current smokers | 3 | 1.53 (1.35, 1.74) | 30% |
Former smokers | 3 | 1.00 (0.92, 1.08) | 0% |
Social isolation | 1 | 1.47 (0.80, 2.70) | - |
TBI | 3 | 1.19 (0.95, 1.50) | 0% |
Vitamin D deficiency | 1 | 1.87 (1.03, 3.40) | - |
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Alcohol consumption | |||
Any versus none | 2 | 0.60 (0.42, 0.85) | 13% |
Light/moderate versus none | 2 | 0.65 (0.56, 0.77) | 60% |
Heavy versus none | 1 | 1.36 (0.68, 2.72) | - |
Heavy versus light/moderate | 1 | 1.89 (1.12, 3.18) | - |
Body weight (mid-life) | |||
BMI ≥30 versus normal | 5 | 2.13 (1.75, 2.59) | 52% |
BMI 25–29.9 versus normal | 3 | 1.40 (1.25, 1.57) | 0% |
BMI <18.5 versus normal | 3 | 1.43 (0.95, 2.17) | 57% |
Body weight (late life) | |||
BMI ≥30 versus normal | 3 | 0.75 (0.67, 0.93) | 4% |
BMI 25–29.9 versus normal | 2 | 0.71 (0.59, 0.84) | 0% |
BMI <18.5 versus normal | 1 | 1.17 (0.70, 1.94) | - |
Body weight (mixed age-groups) | |||
BMI ≥30 versus normal | 3 | 1.30 (0.80, 2.11) | 79% |
BMI 25–29.9 versus normal | 1 | 0.89 (0.81, 0.98) | - |
BMI <18.5 versus normal | 2 | 1.37 (1.20, 1.57) | 6% |
Depression | 6 | 1.64 (1.20, 2.23) | 92% |
Diabetes mellitus | 4 | 1.39 (1.18, 1.65) | 92% |
Diet | |||
Low fish consumption versus high | 2 | 1.39 (1.23, 1.54) | 0% |
Low milk consumption versus high | 1 | 1.54 (1.10, 2.17) | - |
Healthy dietary patterns | 1 | 0.46 (0.27, 0.79) | - |
Hypertension (mid-life) | |||
Hypertension (combined) | 3 | 1.19 (1.08, 1.32) | 0% |
High systolic pressure | 2 | 1.37 (0.97, 1.92) | 0% |
High diastolic pressure | 2 | 1.53 (1.12, 2.10) | 0% |
Hypertension (late life) | |||
Hypertension (combined) | 3 | 0.91 (0.82, 1.02) | 0% |
High systolic pressure | 2 | 0.97 (0.81, 1.16) | 0% |
High diastolic pressure | 2 | 0.79 (0.54, 1.14) | 0% |
Hypertension (mixed age-groups) | |||
Hypertension | 2 | 1.05 (1.04, 1.06) | 0% |
High systolic pressure | 3 | 1.04 (1.01, 1.06) | 75% |
High diastolic pressure | 1 | 1.33 (0.62, 2.86) | - |
Less education | 1 | 1.35 (1.14, 1.60) | - |
Physical inactivity | 2 | 1.56 (1.28, 1.90) | 0% |
Sensory loss | |||
Hearing loss | 2 | 2.03 (1.31, 3.14) | 0% |
Visual impairment | 1 | 1.27 (0.53, 3.04) | - |
Sleep disturbance | 1 | 1.55 (1.25, 1.93) | - |
Smoking | |||
Ever smokers | 2 | 1.12 (1.00, 1.26) | 0% |
Current smokers | 3 | 1.53 (1.35, 1.74) | 30% |
Former smokers | 3 | 1.00 (0.92, 1.08) | 0% |
Social isolation | 1 | 1.47 (0.80, 2.70) | - |
TBI | 3 | 1.19 (0.95, 1.50) | 0% |
Vitamin D deficiency | 1 | 1.87 (1.03, 3.40) | - |
Risk Factors for Vascular Dementia
We extracted results from 29 meta-analyses concerning vascular dementia for 6 risk factors (Table 5). We found statistically significant positive associations between vascular dementia and body weight (BMI ≥30 and BMI 25–29 at mid-life; BMI <18.5 in late life; BMI ≥30 and BMI ≤18.5 in a mixed age-group), depression, hypertension (combined hypertension at mid-life and late life; high systolic pressure at mid-life and late life; high diastolic pressure at late life), and ever or current smoking. The largest risk ratio was observed with combined hypertension at mid-life at 7.68 (3.50, 16.85). Heavy alcohol consumption was not a significant risk factor compared to no alcohol consumption but was a significant risk factor when compared to light/moderate alcohol consumption. Factors that were protective included BMI ≥30 at late life and light/moderate alcohol intake versus none. The I2 statistic could be calculated for 8 associations, all of which were <50%.
Results of meta-analysis of risk factors for vascular dementia
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Alcohol consumption | |||
Light/moderate versus none | 1 | 0.75 (0.57, 0.98) | - |
Heavy versus none | 1 | 0.92 (0.59, 1.44) | - |
Heavy versus light/moderate | 1 | 1.23 (1.03, 1.47) | - |
Body weight (mid-life) | |||
BMI ≥30 versus normal | 2 | 4.04 (2.59, 6.31) | 26% |
BMI 25–29.9 versus normal | 2 | 1.39 (1.12, 1.72) | 0% |
BMI <18.5 versus normal | 1 | 1.47 (0.50, 4.33) | - |
Body weight (late life) | |||
BMI ≥30 versus normal | 2 | 0.73 (0.61, 0.86) | 0% |
BMI 25–29.9 versus normal | 1 | 1.00 (0.74, 1.35) | - |
BMI <18.5 versus normal | 1 | 2.18 (1.18, 4.02) | - |
Body weight (mixed age-groups) | |||
BMI ≥30 versus normal | 2 | 1.49 (1.11, 2.00) | 0% |
BMI 25–29.9 versus normal | 1 | 1.03 (0.84, 1.26) | - |
BMI <18.5 versus normal | 1 | 1.76 (1.20, 2.58) | - |
Depression | 2 | 2.57 (2.16, 3.07) | 0% |
Hypertension (mid-life) | |||
Hypertension (combined) | 1 | 7.68 (3.50, 16.85) | - |
High systolic pressure | 1 | 1.30 (1.08, 1.57) | - |
Hypertension (late life) | |||
Hypertension (combined) | 1 | 3.69 (1.58, 8.70) | - |
High systolic pressure | 1 | 1.40 (1.17, 1.67) | - |
High diastolic pressure | 1 | 4.60 (1.79, 11.83) | - |
Sensory loss | |||
Hearing loss | 1 | 2.26 (0.99, 5.18) | - |
Smoking | |||
Ever smokers | 1 | 1.25 (1.06, 1.48) | 0% |
Current smokers | 2 | 1.51 (1.19, 1.91) | 43% |
Former smokers | 2 | 0.98 (0.84, 1.13) | 0% |
Risk factor . | Number of reviews . | RR (95% CI) . | Heterogeneity, I2 . |
---|---|---|---|
Alcohol consumption | |||
Light/moderate versus none | 1 | 0.75 (0.57, 0.98) | - |
Heavy versus none | 1 | 0.92 (0.59, 1.44) | - |
Heavy versus light/moderate | 1 | 1.23 (1.03, 1.47) | - |
Body weight (mid-life) | |||
BMI ≥30 versus normal | 2 | 4.04 (2.59, 6.31) | 26% |
BMI 25–29.9 versus normal | 2 | 1.39 (1.12, 1.72) | 0% |
BMI <18.5 versus normal | 1 | 1.47 (0.50, 4.33) | - |
Body weight (late life) | |||
BMI ≥30 versus normal | 2 | 0.73 (0.61, 0.86) | 0% |
BMI 25–29.9 versus normal | 1 | 1.00 (0.74, 1.35) | - |
BMI <18.5 versus normal | 1 | 2.18 (1.18, 4.02) | - |
Body weight (mixed age-groups) | |||
BMI ≥30 versus normal | 2 | 1.49 (1.11, 2.00) | 0% |
BMI 25–29.9 versus normal | 1 | 1.03 (0.84, 1.26) | - |
BMI <18.5 versus normal | 1 | 1.76 (1.20, 2.58) | - |
Depression | 2 | 2.57 (2.16, 3.07) | 0% |
Hypertension (mid-life) | |||
Hypertension (combined) | 1 | 7.68 (3.50, 16.85) | - |
High systolic pressure | 1 | 1.30 (1.08, 1.57) | - |
Hypertension (late life) | |||
Hypertension (combined) | 1 | 3.69 (1.58, 8.70) | - |
High systolic pressure | 1 | 1.40 (1.17, 1.67) | - |
High diastolic pressure | 1 | 4.60 (1.79, 11.83) | - |
Sensory loss | |||
Hearing loss | 1 | 2.26 (0.99, 5.18) | - |
Smoking | |||
Ever smokers | 1 | 1.25 (1.06, 1.48) | 0% |
Current smokers | 2 | 1.51 (1.19, 1.91) | 43% |
Former smokers | 2 | 0.98 (0.84, 1.13) | 0% |
Data Availability by Type and Etiology of NCD
Table 6 contains a summary of findings across all risk factors and outcomes. It shows risk factors for which at least one statistically significant association was noted for each type and etiology of NCD and risk factors for which no data were available. Notably, we extracted no data on 9 of 14 risk factors for mild NCD and no data on 8 of 14 risk factors for vascular dementia.
Discussion
We conducted a systematic review and meta-analysis of systematic reviews for modifiable risk factors for mild and major NCDs, including an examination across two of the more common etiologies of major NCD. We synthesized findings on 14 broadly defined risk factors, all of which were found to be significantly associated with major NCD. Five of these risk factors were definitely associated with mild NCD based on the available data. In general, the risk estimates were consistent across types and etiologies of NCDs, though some notable differences in magnitude were observed. Far less research has been reported on vascular dementia and mild NCD compared to major NCD and AD dementia.
The epidemiology of modifiable risk factors for dementia is a topic of considerable interest as a globally aging population is increasing the burden of mortality and morbidity related to NCDs. Notably, the Lancet Commission produced several reports on risk factors for dementia, which were widely cited by public health agencies, associations, and patient groups [11, 12]. Our umbrella review examined several potentially modifiable risk factors not included in the latest Lancet Commission report (i.e., sleep disturbances, diet, visual impairment, vitamin D deficiency). We found they were significantly associated with a heightened risk of major NCD. These additional factors are important to consider as dementia prevention strategies evolve. Conversely, we do not report on air pollution, which was designated by the Lancet Commission as an important risk factor [70]. While the Lancet Commission only considered major NCD, we also examined two common dementia etiologies and mild NCD. Nevertheless, the results from our review are largely concordant with the Lancet Commission, although we found lower RR estimates for social isolation and a higher for less education. Methodological differences likely underlie these disparities. The Lancet Commission did not utilize an umbrella review methodology, but rather relied on a mix of previous reports and systematic and non-systematic reviews to produce the RRs used in their report. Variation in the definition of risk factors may also play a role in the heterogeneity seen between and within reviews.
We synthesized risk factors estimates on two of the most common etiologies of dementia: AD and vascular disease. While evidence suggests that the majority of individuals with major neurological disorders have co-occurrence of multiple pathologies [71], we were not able to include mixed dementia as a separate subcategory. There was considerably more research on the risk factors for AD dementia than vascular dementia, where we were only able to find reviews of six risk factors. The limited evidence available for vascular dementia is likely a result of the lower incidence of dementia that is felt to be solely due to vascular disease. Even when evidence was available for extraction, there were fewer studies per risk factor and fewer participants per study than with the major NCD reports, reducing the precision of our estimates. Nevertheless, we found that the risk associated with mid-life obesity and hypertension in our study was notably higher for vascular dementia than for AD dementia. We also found that TBIs had a strong association with major NCD, but not with AD dementia, despite the fact that TBIs are considered by many as a risk factor for AD. It is possible post-TBI dementia cases are being categorized as due to TBI, chronic traumatic encephalopathy, or another cause rather than AD.
Our inclusion of mild NCD was a novel element of this study. Mild NCD had the least evidence available for synthesis but was significantly associated with each of the five risk factors where data could be extracted. The magnitude of the risk factors estimates for mild NCD were not consistently different than major NCD with the exception of sleep disturbances where the risk ratio for major NCD was notably higher than mild NCD. Our research question and search strategy were designed to allow us to find risk factors for mild NCD that were not risk factors for major NCD. While we did not find any such factors, we cannot conclude they do not exist. More research is needed on mild NCD given its place on the continuum of neurodegenerative diseases. There is a trend toward developing novel therapeutics for patients with this stage of cognitive loss because it is hoped those treated earlier may be more responsive to disease-modifying interventions for NCDs [72].
Finally, we found imbalances in the extent of research done on specific risk factors. Body weight and depression had by far the most research available to synthesize. While clearly important risk factors, the relative volume of the literature on them may well not be congruent with their overall impact on NCD occurrence. For example, the Lancet Commission report identified that among the risk factors included, less education and hearing loss had the highest population attributable fractions for worldwide dementia cases [11]. These factors and others with more limited evidence such as diet and social isolation should be the focus of further research. Additionally, more research should focus on the implications that the high prevalence of mixed pathologies has for dementia prevention.
Limitations
Our study is limited to English language articles. Furthermore, while there were sufficient data to report body weight and hypertension by life stage, this could not be done for other variables, likely contributing to heterogeneity across reviews. We were unable to explore other sources of heterogeneity, for example, varying definitions of outcomes and risk factors. A limitation of an umbrella review is that it may not include recent primary studies that have not yet been incorporated into a systematic review. As noted previously, we did not examine air pollution, other environmental and occupational risk, or early life adversity. Additionally, as there is not full agreement among researchers on which risk factors are modifiable, there is an element of subjectivity in our determination of which risk factors to include. Finally, we were not able to separately analyze risk factors for other common causes of dementia such as Lewy body, frontotemporal, and mixed dementias.
Conclusion
We conducted an umbrella review and synthesized evidence on 14 modifiable risk factors for major and mild NCDs: alcohol consumption, body weight, depression, diabetes mellitus, diet, hypertension, less formal education, physical inactivity, sensory loss, sleep disturbance, smoking, social isolation, TBI, and vitamin D deficiency. We found evidence that all of these factors are related to the risk of major NCD. Prevention strategies should consider approaches that reduce the incidence and severity of these risk factors through health promotion, identification, and management, using individual and organizational-focused implementation strategies [73]. More research is needed specific to mild NCD and non-Alzheimer’s dementia.
Acknowledgment
We would like to thank Denise Smith for assistance in developing our electronic search strategy.
Statement of Ethics
An ethics statement is not applicable because this study is based exclusively on published literature.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Funding Sources
This study received funding from the Public Health Agency of Canada (2021-HQ-000076). PR holds the Raymond and Margaret Labarge Chair in Optimal Aging and Knowledge Application for Optimal Aging, is the director of the McMaster Institute for Research on Aging and the Labarge Centre for Mobility in Aging, and holds a Tier 1 Canada Research Chair in GeroScience. LG is supported by the McLaughlin Foundation Professorship in Population and Public Health. DD is funded with a doctoral scholarship through the Canadian Institutes of Health Research (CIHR) (funding reference number #FBD-181577). TLA holds a Tier 1 Canada Research Chair in Healthy Aging.
Author Contributions
A.J., M.U.A., M.K., A.M., M.O., T.L.-A., V.T., J.M.M., D.B.H., S.K., A.P.C., C.W., P.R., and L.G. contributed to the conception/design of the work. A.J., M.U.A., M.K., A.M.,V.M.,H.H.,S.L.,E.Y.,K.P.,D.S.,R.D.,K.A.,R.H.C.,D.D., and D.R.M. contributed to data acquisition. M.U.A. conducted data analysis. A.J. drafted the initial manuscript. All authors reviewed the work critically for important intellectual content and approved the final version.
Data Availability Statement
All data used in this study were taken from the published literature. Further inquiries can be directed to the corresponding author.