“The number of people living with dementia worldwide in 2019 was estimated at 57 million and is projected to increase to 153 million by 2050. The proportion of people with dementia has increased over time in lower-income countries due to a greater percentage increase in longevity than in high-income countries. [..]
There has been a rapid expansion in the volume of work on dementia prevention and risk reduction related to the 12 risk factors that were identified from the existing research literature and discussed in our earlier Lancet Commission reports in 2017 and 2020. The risk factors identified in our earlier reports were less education, hearing loss, hypertension, physical inactivity, diabetes, social isolation, excessive alcohol consumption, air pollution, smoking, obesity, traumatic brain injury, and depression—for which we reported that reductions have the potential to prevent 40% of cases of dementia. We discussed the mechanisms for these 12 risk factors, which indicated that risk can be reduced at any age. [..]
People with healthy lifestyles, involving regular exercise, not smoking, avoiding excess alcohol, and including cognitive activity in late life, were shown not only to have a lower risk of dementia than those with less healthy lifestyles but also to have dementia onset delayed for longer than their increased life expectancy, resulting in more healthy years and fewer years of illness. Overall, people living healthier lives can expect to live longer than people with unhealthy lifestyles, and if they develop dementia, live fewer years with the disease, with notable quality-of-life implications for individuals and cost-saving implications for services. [..]
[Education]
We previously reported that people with more childhood education and higher educational attainment have a reduced dementia risk, and discussed whether the effects of later cognitive stimulation might be due to people with more education having more cognitively stimulating occupations than people with low levels of education. Differences in the quality of education, as measured by reading levels at ages 14–15 years, have been estimated to account for about half of the US disparities in dementia prevalence across racial groups. Overall, educational attainment, not years of education, appears to drive the protective effect for future cognition and dementia. [..]
[Hearing Loss]
None of these analyses included all of the criteria that we judged to ensure high-quality data in our previous meta-analysis. We also excluded studies comparing populations with varying severities of hearing loss, but not comparing individuals with hearing loss with those without hearing loss. We searched again from database inception until March 20, 2023, on PubMed, Ovid Embase, PsycINFO, Web of Science, Cochrane Library, PROSPERO, and the Centre for Reviews and Dissemination, contacting authors for clarification as needed, and found six studies fitting the criteria. We searched “all fields” using the search terms “dementia” or “cognitive decline” or “Alzheimer’s disease” or “mild cognitive impairment” AND “hearing” or “auditory” or “aural” or “presvycusis”. We calculated totals if only subgroups were reported, generating an overall HR for studies. We used results unadjusted for hearing aids because hearing aids are part of the causal pathway between hearing loss and dementia. The mean baseline age of study participants ranged from 59 years to 77 years, with the largest study recruiting men when they enrolled in the mandatory conscription board at age 18–20 years but measuring hearing status at a median age of 59·9 years (IQR 54·6–65·4). Follow-up in all studies between baseline and dementia status was between 6 years and 12 years. We conducted a random-effects meta-analysis of these studies, in which people with hearing loss had an increased risk of dementia compared with those without hearing loss (HR 1·37, 95% CI 1·00—1·87; I2=80%; n=666 370). Four of the smaller studies reported hearing aid use, and 18–64·5% of people with hearing loss wore hearing aids. All people with hearing loss were included in our meta-analysis, without considering the use of hearing aids in the overall risk estimate, so the estimate is conservative. In our meta-regression, studies with a higher proportion of people who wore hearing aids reported a lower likelihood of dementia than those with a lower proportion of people who used hearing aids, but the confidence interval was wide (–1·32, –3·34 to 0·71). [..]
The evidence described here raises the question of whether the use of hearing aids in people with hearing loss can eliminate or mitigate the increased dementia risk. The ACHIEVE study, the first RCT of hearing aids and cognition, recruited people aged 70–84 years. The participants were healthy volunteers with hearing loss who were recruited with advertisements (n=739) and people from an existing cohort, the ARIC study (n=238). There was no overall effect of the use of hearing aids on the primary outcome of cognition at 3-year follow-up (difference –0·002, 95% CI –0·08 to 0·08). Importantly, a prespecified sensitivity analysis identified substantial effects of hearing aid use on cognition at 3 years in the ARIC group (difference 0·19, 0·02 to 0·36). The ARIC population had more risk factors for dementia (ie, the mean population age was 2·8 years older, lower baseline cognition, smoked more, less education, more often lived alone, and more likely to have diabetes and hypertension) than the healthy volunteer population with hearing loss. Incident cognitive impairment was higher in the ARIC group (57 [24%] of 238 participants) than in people who were recruited via advertisements (61 [8%] of 739 participants) at 3-year follow-up. Notably, the authors emphasised that volunteer participants who are recruited through this type of method generally represent a healthier subset of the target population. Overall, there was a large protective effect of hearing aids on cognition in the population at high risk in the ARIC cohort (48% reduction in 3-year global cognitive decline compared with the control population). The slower rate of cognitive decline in the healthy volunteer group compared with the ARIC cohort might have limited any effect on cognition in this group within a 3-year follow-up period. The explanation of the large effect in the ARIC cohort might be that hearing aids in groups at high risk of dementia also change social contact, low mood, cognitive stimulation, and improve motivation and communication about medical treatment, but this evidence does not exist yet.
We previously discussed the evidence that hearing aid use is protective against dementia and reduces cognitive deterioration rates after beginning hearing aid use. Since then, a systematic review and meta-analysis of eight cohort studies with 126 903 participants, followed up for 2–25 years, reported that people with hearing loss who used hearing aids had a significantly lower risk of cognitive decline (HR 0·81, 0·76–0·87; I2=0%) and dementia (0·83, 0·77–0·90; I2=0%; four studies) than those who did not use assistive devices.
In another cohort of 2114 people older than 50 years with self-reported hearing loss, 1154 people had MCI and those that used hearing aids were at significantly lower risk of developing all-cause dementia during the follow-up than those not using hearing aids (HR 0·73, 0·61–0·89). The median time to incident dementia was 2 years for non-hearing-aid users and 4 years for hearing aid users.
[Smoking]
A large meta-analysis by Zhong and colleagues reported that midlife smoking increased dementia risk (RR 1·30, 95% CI 1·18–1·45; 37 studies) but there was no increased risk in former smokers. The Framingham Heart Study (n=4015; 21-year follow-up) identified the strongest risk for dementia in people who started smoking in early adult life (ie, aged 33–44 years; HR 1·42, 95% CI 0·05–3·60). Other long-term cohort studies, such as the ARIC study (25-year follow-up; n=15 744; 1·41, 1·23–1·61) and the Whitehall II study (32-year follow-up; n=9951; 1·36, 1·10–1·68), have reported similar excess risks of dementia in midlife (ie, mean age 44·9 years [SD 6·0]) current smokers. A UK Biobank study of 497 401 adults reported an HR for dementia of 1·7 (1·2–2·5) for smokers younger than 50 years at baseline. In the Danish general population, a pooled analysis of two prospective cohorts, including a total of 61 664 individuals, reported that risk of dementia for midlife smokers was increased for men (3·2, 1·4–7·4) and women (1·7, 1·1–2·8) compared with non-smokers.
A 32-year follow-up of the Whitehall II cohort, controlling for socioeconomic status, identified that current smokers (HR 1·36, 95% CI 1·10–1·68) but not ex-smokers (0·95, 0·79–1·14) have an increased risk of dementia compared with people who have never smoked and that socioeconomic inequalities in dementia risk were partly mediated by smoking. The study by Zhong and colleagues also showed no increased risk in former smokers. Similarly, a Korean nationwide study of 789 532 participants who were assessed for smoking status over 2 years reported that ex-smokers had a lower risk of all cause dementia (0·92, 0·87–0·97) than continuing smokers, which was more pronounced among adults who smoked before age 65 years (0·8, 0·7–0·9) than those who smoked at age 65 years or older (1·0, 0·9–1·0). Another Korean population study examining dementia risk in people with atrial fibrillation also reported a reduced risk of dementia in people who had quit smoking (0·83, 0·72–0·95) compared with current smokers. These studies suggest that smoking cessation reduces dementia risk compared with continued smoking. Smoking should now be considered a midlife risk factor (rather than a late-life factor, as in the 2020 Lancet Commission), and the beneficial effect of stopping smoking is encouraging.
[Cardiovascular Risk Factors]
Vascular dementia usually occurs when people have a stroke (and stroke is part of the diagnostic criteria), and vascular dementia happens more often in people who smoke or who have diabetes and hypertension. Stroke and dementia share the risk factors of less education, infrequent exercise, hypertension, heart disease, and social isolation, but some people with these risk factors will not develop dementia despite neuropathology, sometimes because they die at a young age, before dementia develops.
Several studies have examined the effect of a combination of cardiovascular risk factors on dementia risk. The Life’s Simple 7 group defined ideal cardiovascular health factors (ie, BMI, diet, smoking, physical activity, blood pressure, cholesterol, and glucose concentrations [The American Heart Association has since added sleep to its cardiovascular health factor list, making it “Life’s Essential 8.” Healthy sleep is defined as 7-9 hours of sleep each night for most adults.]), and higher scores on this index are associated with lower dementia risk. Similarly, in China, a 10-year longitudinal study of 29 072 people with mean age of 72 years at follow-up (SD 6·6) reported that slow memory decline was associated with being in the healthy group, which meant having at least four of six factors: healthy diet (ie, eating at least seven of 12 eligible food items), physical exercise (ie, ≥150 min of moderate intensity or ≥75 min of vigorous intensity exercise weekly), active social contact (ie, ≥2 social contacts per week, including online), active cognitive activity (ie, engaging in ≥2 cognitive activities per week), never or previously smoking (ie, quit smoking ≥3 years ago), and never drinking alcohol (ie, drinking less than a small glass of wine daily). This association applied to both APOE ε4 carriers and non-carriers.
[LDL cholesterol]
[..] a meta-analysis of three cohort studies with a total of 1 138 488 participants, all from the UK, looking at LDL cholesterol in adults younger than 65 years followed up for more than 12 months reported that each 1 mmol/L increase in LDL cholesterol was associated with an 8% increase in incidence of all-cause dementia (effect size 1·08, 95% CI 1·03–1·14; I2=0·3%). A study of 1 189 090 participants reported that high LDL cholesterol (ie, >3 mmol/L) was associated with an increased risk of dementia (HR 1·33, 95% CI 1·26–1·41). In a large cohort from the UK Clinical Practice Research Datalink (n=1 853 954) who were followed up for a median of 7·4 years (IQR 4·6–10·4), higher baseline LDL cholesterol was similarly associated with increased risk of all-cause dementia (adjusted rate ratio 1·05, 95% CI 1·03–1·06 per SD increase in LDL cholesterol [ie, 1·01 mmol/L or 39 mg/dL increase]). This risk was stronger in people younger than 65 years at baseline for dementia diagnosed within 10 years (1·10, 1·04–1·15) and more than 10 years after baseline (1·17, 1·08–1·27) than for people who were older than 65 years at baseline. In a Danish cohort study of 94 184 people followed up from a mean age of 58 years (SD 13·0), people who did not adhere to dietary guidelines (ie, eat at least three weekly servings of all of fruit, vegetables, and fish; rarely drink sugar-sweetened drinks; rarely eat prepared meat like sausages or have takeaways) were more likely to have high LDL cholesterol. After a median follow-up of 9 years (range <1–15), people with low adherence to these guidelines were more likely to develop dementia types other than Alzheimer's disease than were people with high adherence (HR 1·54, 95% CI 1·18–2·00), but were not more likely to develop Alzheimer's disease, although subtyping of dementia might not have been accurate. People who took lipid-lowering drugs did not have an increased risk of dementia. A US study of 4392 people reported that increased HDL cholesterol protected against the development of dementia. Further evidence of causality comes from a mendelian randomisation meta-analysis that included 27 studies, including 3136 people with dementia and 3103 healthy controls, which reported that high total cholesterol and low HDL cholesterol were risk factors for dementia. By contrast, an individual participant meta-analysis of more than 21 000 people (mean baseline age 76 years) identified no association between total cholesterol, LDL cholesterol, or HDL cholesterol and cognitive decline. This result did not change when the analysis was stratified by statin use or APOE ε4 status. Excess brain cholesterol is associated with increased stroke risk and deposition of brain amyloid β and tau, suggesting a potential mechanism for the link between LDL cholesterol and dementia. HDL reduces excessive cholesterol and is inversely correlated with brain amyloid β concentration. Individual counselling about diet and exercise has a small effect in reducing LDL cholesterol. Statins have become a focus of research in the field of Alzheimer's disease and have potential benefit due to their anti-inflammatory and antioxidant properties as well as reducing cholesterol. A meta-analysis of 36 cohort studies identified that statin use was associated with a reduced risk of all-cause dementia (OR 0·80, 95% CI 0·75–0·86; I2=97·5%) and Alzheimer's disease (0·68, 0·56–0·81; I2=94·5%) compared with untreated high cholesterol, with no difference between men and women. A Cochrane review of RCTs of statins given in late life found no effect on either dementia risk (one study) or cognitive outcomes (two studies). Repeat observational data can be used to emulate a target trial of statin use. By use of data from 6373 participants aged 55–80 years, an emulated trial identified that sustained statin use, but not statin initiation alone, was associated with reduced 10-year risk of dementia or death. Overall, high-quality, consistent, biologically plausible evidence exists that high LDL cholesterol in midlife is a risk factor for dementia. The 2019 WHO guidelines suggested that management of dyslipidaemia in midlife could be offered to reduce the risk of cognitive decline and dementia but that the quality of evidence was low. Although long-term, high-quality RCTs of statins to prevent dementia do not exist, these studies would be unethical and impractical to run. Meta-analyses of observational studies are heterogenous but show benefit of statins on dementia risk, possibly because the benefit depends on age of initiation.
[Physical inactivity, exercise, and fitness]
[..] Since the 2020 Lancet Commission, a systematic review and meta-analysis of 58 studies (n=257 983) exploring the link between physical activity and dementia identified that physical activity was associated with a decreased risk of all-cause dementia (RR 0·80, 95% CI 0·77–0·84) and Alzheimer’s disease (0·86, 0·80–0·93) in short and long follow-ups of at least 20 years, regardless of baseline age. There was decreased risk of vascular dementia in shorter follow-ups (0·79, 0·66–0·95) of a mean of 10·9 years (SD 8·5). A range of intensities of exercise were included in the meta-analysis, and reduction in risk was greatest when moving from extreme sedentariness to some physical activity. A cohort study (n=1417) that recorded physical activity five times between ages 36 years and 69 years reported that being physically active at all ages was associated with better cognition at age 69 years than no physical activity, with the strongest association for sustained physical activity. A prospective study of 29 826 people who were followed up for a median of 24·5 years (IQR 24·1–25·0) and whose weekly physical activity was assessed twice, 10 years apart, reported that people who maintained an individually estimated optimal level of physical activity had a reduced risk of dementia compared with persistently inactive individuals (HR 0·75, 95% CI 0·58–0·97), as did those who increased their physical activity to an optimal level (0·83, 0·72–0·96). A longitudinal study of 1718 women over a median of 11·9 years (range 0·6–13·5) reported that higher physical activity level was associated with less cognitive decline, but not when the estimate was adjusted for diabetes and hypertension.
In an RCT, 945 participants, with a mean age of 78 years (SD 2) and 450 (48%) women and 495 (52%) men, were randomly assigned (2:1:1) to a 5-year control group, moderate-intensity continuous training twice a week, or high-intensity interval training twice a week. At 5 years, 474 (96%) of 494 participants in the control group adhered to national guidance for physical activity, 176 (75%) participants adhered to the moderate-intensity interval training intervention, and 164 (76%) adhered to the high-intensity interval training intervention. There was no significant difference in cognition (β 0·26, 95% CI −0·17 to 0·69) or odds of MCI (OR 0·86, 95% CI 0·66 to 1·13) between the groups. Men in the combined moderate or high intensity exercise group had a decreased risk of MCI (0·68, 0·47 to 0·99) and slightly higher cognitive scores than male participants in the control group. Participants who decreased their peak oxygen uptake, rather than maintained or increased their uptake, had increased odds of MCI (1·35, 0·98 to 1·87) compared with those with stable oxygen uptake levels, although this result was imprecisely estimated. Findings are in line with the small cognitive benefit shown in an umbrella review of RCTs on the effects of physical exercise on cognition. Outcomes might depend on not only the duration but also the type and intensity of physical activity.
[Diabetes]
[..] New evidence suggests that age of onset makes a difference, with midlife, but not necessarily late-life, diabetes onset increasing the risk of dementia. In a prospective cohort study of 10 095 participants, the risk for dementia increased for every 5-year decrease in age of type 2 diabetes onset (HR 1·24, 95% CI 1·06–1·46), until aged over 70 years at onset. Diabetes should be classified as a midlife risk for dementia. It is unclear whether diabetes is not a risk factor for dementia at older ages or whether the absence of evidence showing significant risk is because of short follow-ups and few studies. [..] Long illness duration and poorly controlled diabetes increase the risk of dementia.
Our understanding of the mechanism by which diabetes increases the risk of dementia is incomplete. Long-term microvascular and macrovascular complications are well established in diabetes, and the causal mechanism likely incorporates a strong vascular component, including stroke risk. Peripheral insulin resistance leads to decreased insulin signalling in the CNS, followed by alteration in brain metabolism. Insulin resistance is a common molecular mechanism linking diabetes and Alzheimer’s disease: it leads to increased amyloid β toxicity, tau hyperphosphorylation, oxidative stress, and neuroinflammation. Increased concentrations of systemic inflammatory markers (eg, CRP) were associated with the diabetes-associated increased dementia risk.
It is unclear whether effective treatment of diabetes ameliorates dementia risk per se, particularly as taking large quantities of oral medication and insulin is related to increased severity of diabetes. Strict, intensive treatment compared with standard diabetic control, however, does not decrease the risk of dementia. Some evidence suggests that people taking some types of anti-diabetic medication might be less at risk of dementia. A systematic review, meta-analysis, and network analysis of 27 studies (1 590 757 participants), which did not report heterogeneity, identified that cohort studies indicated that SGLT2 inhibitors (OR 0·41, 95% CI 0·22–0·76), GLP-1 receptor agonists (0·34, 0·14–0·85), and DPP-4 inhibitors (0·78, 0·61–0·99) were associated with dementia risk reduction, whereas sulfonylureas were associated with increased risk (1·43, 1·11–1·82). Metformin was not associated with a decreased or increased risk (0·71, 0·46–1·08). A study in UK primary care reported a significantly lower risk of dementia in 114 628 people with diabetes initiating metformin than in 95 609 people who were not on medication for their diabetes (HR 0·88, 95% CI 0·84–0·92). A meta-analysis of three RCTs and a cohort study of GLP-1 receptor agonists in people with type 2 diabetes identified a lower dementia rate in people who were randomly assigned to the drug than in those on placebo (15 820 participants; HR 0·47, 95% CI 0·25–0·86) and in people on GLP-1 receptor agonists in the nationwide Danish cohort (120 054 individuals; 0·89, 0·86–0·93). Another meta-analysis of observational studies of 819 511 people with type 2 diabetes and a mean follow-up of 4·5 years (range 1·3–7·2) reported similar findings, with less subsequent dementia in users of SLGT2 inhibitors (three studies; RR 0·62, 95% CI 0·39–0·97; I2=82·5%), GLP-1 receptor agonists (four studies; 0·72, 0·54–0·97; I2=91·3%), and DPP-4 inhibitors (seven studies; 0·84, 0·74–0·94; I2=88·6%) than in people not using these medications, but the analysis reported high heterogeneity. People with diabetes might not be taking medication because their diabetes is well controlled without medication or because it is not well treated, which might account for the heterogeneity between studies. Evidence from RCTs also exists for the protective effect of GLP-1 receptor agonists. In a Taiwanese population of 31 384 propensity-matched pairs (including matching for chronic kidney disease with diabetes) who were followed up for 5 years, people who were adherent to metformin had a 72% lower risk of developing dementia than people who did not adhere.[..]
Weight loss might also help to control diabetes and therefore might also affect cognition. The Look AHEAD study recruited 3751 people aged 45–76 years with type 2 diabetes and overweight or obesity and randomly assigned them to a 10-year intervention of increased exercise and decreased calorie intake or diabetes support and education. This RCT was terminated because the interim analysis showed that the intervention had no effect on death from vascular outcomes or myocardial infarction, stroke, or severe angina. Cognitive outcome was measured at follow-up, controlling for baseline education but not cognition. There was a strong inverse relationship between HbA1c concentration and cognition over both groups. Cognitive function was not related to group allocation or to weight loss. Overall improved control of diabetes, but not very low blood sugar or weight loss without improved diabetic control, might attenuate the risk of dementia and be a way of decreasing dementia risk.
[Hypertension and its trajectory]
[..] There are three meta-analyses of RCTs for antihypertensive medication. Two meta-analyses identified that antihypertensive medication was protective against cognitive impairment and dementia, and one with short (ie, range 1–5 years) follow-up did not identify a protective effect. The meta-analysis of 12 RCTs (n=96 158), with a mean follow-up of 4·1 years (range 2·2–5·7), identified a lower risk of dementia or cognitive impairment in people taking antihypertensive medications than in controls, consisting of people taking placebo, taking alternative antihypertensive agents, or whose target blood pressure was higher than that of the intervention group (OR 0·93, 95% CI 0·88–0·98), and of cognitive impairment alone (0·93, 0·88–0·99). The second meta-analysis used individual participant data from five RCTs (n=28 008) with placebo controls, three of which were included in the first meta-analysis, and identified a lower risk of dementia in the treatment group than in the placebo control group (0·87, 0·75–0·99). A Cochrane review, with three studies overlapping with the previous meta-analysis, included 12 RCTs (eight placebo-controlled trials; n=30 412) with interventions lasting at least 12 months. The review concluded that there was a modest benefit of antihypertensive medication on cognitive change measured with the Mini Mental State Examination (MMSE; five studies; mean difference 0·20, 95% CI 0·10–0·29), but duration was too short to show a difference in dementia incidence (four studies; OR 0·89, 95% CI 0·72–1·09). An individual participant data meta-analysis of 17 studies including people in LMICs and HICs (mean age 72·5 years [SD 7·5], follow-up 4·3 years) identified that people with untreated hypertension had a higher risk of dementia than healthy controls (HR 1·42, 95% CI 1·15–1·76), but this risk was attenuated or lost with treatment (1·13, 0·99–1·28). One meta-analysis of individual participant data cohorts comprising 31 090 adults without dementia at baseline, with a follow-up of at least 5 years, found that people with hypertension who were taking any antihypertensive drug were at lower risk of dementia than those who were not taking antihypertensive drugs (HR 0·88, 95% CI 0·79–0·98), but did not identify any differences between classes of drugs. Although there is a scarcity of direct comparisons of the effect of different antihypertensives, a network analysis and systematic review identified that treatments with angiotensin 2 receptor blockers and calcium channel blockers (CCBs) were associated with lower dementia risk than other antihypertensives.
[Obesity and weight]
[..] A further systematic review and meta-analysis examining the association between obesity and dementia included 14 studies with 77 890 participants and identified that midlife obesity was associated with subsequent all-cause dementia (RR 1·31, 95% CI 1·02–1·68). Another study on central obesity, measured through waist circumference or waist-to-hip ratio, included 5 060 687 participants from 16 studies and showed that larger versus smaller waist circumference was associated with a greater risk of cognitive impairment and dementia (HR 1·10, 95% CI 1·05–1·15), and this risk was greater in people older than 65 years than other ages. Obesity is more common in people who exercise infrequently and is associated with diabetes and hypertension, which also cause cardiovascular disease, so this association could possibly be mediated by other risk factors for dementia. Nonetheless, most studies in these meta-analyses adjusted for health conditions, such as hypertension, stroke, blood lipid concentrations, and diabetes, as well as demographic characteristics, which should have minimised the effect of these intermediaries.
A meta-analysis of interventional studies for weight loss identified 13 longitudinal studies (n=551) and seven RCTs (n=468) of participants with overweight (ie, BMI 25·0–29·9 kg/m2) or obesity (ie, BMI ≥30·0 kg/m2). Intentional modest weight loss of even 2 kg among trial participants was associated with improvements in cognition at median follow-up of 6 months (range 8–48),209 indicating that health behaviours could have a beneficial effect, even if weight loss is not sufficient to alter obesity status. These improvements were more pronounced in people who changed their diet or who exercised to lose weight than in people who had bariatric surgery. [..]
A systematic review of 19 prospective studies, in which data were pooled, also identified an increased risk of dementia in people who were underweight (ie, BMI <18·5; HR 1·26, 95% CI 1·20–1·31). A meta-analysis of individual participant data from 1·3 million people across 39 prospective cohort studies found that obesity was a risk factor for dementia in cohorts where the baseline measurement was taken more than 15 years before dementia onset but appeared to be protective if the baseline measurement was taken less than 10 years before dementia onset. The authors suggested that this result was due to reverse causation, because people often lose weight before they develop dementia. Being underweight is also potentially linked with malnutrition, although being underweight can occur for many reasons.
[Excessive alcohol consumption]
[..] individual participant meta-analysis of 131 415 participants from France, the UK, Sweden, and Finland found that, after adjusting for confounders, heavier drinking (ie, >21 units per week) in midlife was associated with an increased risk of dementia compared with lighter drinking (ie, 1–21 units per week; HR 1·22, 95% CI 1·01–1·48). In line with this finding, a review of 28 systematic reviews concluded that heavy alcohol use (as defined by the individual studies) was associated with an increased risk of all-cause dementia and reduced grey matter volume in imaging studies. Alcohol-induced loss of consciousness increased dementia risk in people with either moderate or heavy consumption.
[..] A Japanese prospective study followed up 42 870 participants for 14·9 years and reported that not drinking (HR 1·29, 95% CI 1·12–1·47) and drinking more than 450 g of alcohol per week from midlife (1·34, 1·12–1·60) were associated with increased risk of dementia compared with light drinking (ie, <75 g/week). A meta-analysis of individual participant data from 24 478 older adults (mean age 71·8 years [SD 7·5]) across 15 prospective cohort studies reported that, during 151 636 person-years of follow-up, dementia risk was lower in occasional (0·78, 0·68–0·89), light-to-moderate (1·3–24·9 g/day; 0·78, 0·70–0·87), and moderate-to-heavy drinkers (25·0–44·9 g/day; 0·62, 0·51–0·77) than in non-drinkers but was not lower in heavy drinkers (>45 g/day) than in non-drinkers (0·81, 0·61–1·08). Mendelian randomisation also indicated a causal link between alcohol consumption and an earlier age of onset of Alzheimer’s disease and suggested that any link between not drinking and Alzheimer’s disease is due to survivor bias. Observational studies usually find a J-shaped dose–response, such that not drinking is associated with increasing dementia risk compared with light drinking. This result is probably because many non-drinkers have previously had high alcohol consumption or other illnesses that prevent them from drinking, and studies that correct for previous high alcohol consumption have reported that there is no excess mortality in the non-drinking group.
A study of a nationwide, South Korean cohort of 3 933 382 participants that serially assessed alcohol consumption over 3 years reported that sustained heavy drinkers (ie, ≥30 g/day or 3 units per day) had an increased risk of all-cause dementia (HR 1·08, 95% CI 1·03–1·12), and reducing drinking from heavy to moderate levels (ie, 15·0–29·9 g/day) reduced the risk of all cause dementia (0·92, 0·86–0·99) compared with sustained heavy drinking.220 Sustained mild (ie, <15 g/day; 0·79, 0·77–0·81) or moderate alcohol consumption (0·83, 0·79–0·88) or initiating mild alcohol consumption (0·93, 0·90–0·96) were also associated with lower risk of all-cause dementia than sustained non-drinking; however, some non-drinkers might have been former heavy drinkers. Overall, reduction of excessive alcohol or sustained light drinking is associated with a lower dementia risk than is excessive alcohol. A lack of clear evidence exists that not drinking alcohol increases the risk of dementia. The observational evidence of excess risk for non-drinkers might be due to people who have previously drunk large amounts, abstained at the time when data were gathered (and been classified as non-drinkers), and then might return to drinking.
[Social isolation]
[..] two systematic reviews reported that less frequent social contact was associated with higher risk of dementia. The first review, which included eight studies with a total of 15 762 participants, reported higher dementia risk (RR 1·57, 95% CI 1·32–1·85) for people with less frequent social contact than for people with more frequent social contact, with social contact dichotomised as less or more frequent within individual studies and the amount of contact varying between different studies. The second review (which included one study that was included in the previous review) reported a smaller increased risk (1·18, 1·08–1·30). Duration of follow-up might partly explain the inconsistent results from these studies; for example, seven of the eight studies in one of the reviews had less than 4 years of follow-up, making reverse causation likely. However, two subsequent studies of participants from the UK Biobank, with mean follow-up of 8·8 years223 and 12 years (SD 1·7), found that dementia risk was higher in people who were more socially isolated (ie, defined as meeting at least two of three criteria of: living alone, seeing family or friends less than once a month, and participating in no weekly group activities) at baseline than less socially isolated individuals (ie, people who did not meet these criteria).
Loneliness is linked to, but differs from, social isolation because it is about people’s feelings that their social contact is inadequate. Loneliness was also associated with increased dementia risk in three reviews comprising three to eight studies (RR 1·26, 1·14–1·40; 1·38, 0·98–1·94; and 1·58, 1·19–2·09). An increased dementia risk of 34–91% was reported in subsequent studies, in the USA over 10 years, in the Netherlands and Sweden over 14 years, and in Japan over 5 years. Some, but not all, of these studies found that the association persisted after adjustment for potential confounders, including infrequent social contact.
[..] Interventions to increase social contact and participation in activities through facilitator-led group activities have yielded inconsistent results on general cognitive function. One Finnish RCT of a 3-month intervention with a primary outcome of cognition recruited 235 people aged 75 years or older who were lonely and showed a small significant improvement in performance on the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (mean difference in change of –2·6 points per 100 points), but studies from the USA and China did not show that facilitator-led group activities were beneficial. Studies of multidomain interventions that included group components suggested small cognitive benefits (Cohen’s d 0·13; mean MMSE difference of 0·99 points) for highly intensive interventions. A subsequent pilot RCT of a multidomain intervention, including social activities through group meetings and additional scheduled monthly social activities, led to general cognitive improvement at 24 weeks despite small numbers (between-group difference of 6·2 points on the Repeatable Battery for the Assessment of Neuropsychological Status score; p=0·004). The contribution of the social component of multidomain intervention is unclear. Existing studies are too small and follow-up is too short to identify whether these social components of multidomain interventions have any effect on dementia incidence.
[Air pollution]
[..] Continuing research interest is reflected by the publication of at least nine further systematic reviews and meta-analyses since 2019, which have all reported that air pollution is associated with increased dementia risk. To manage study heterogeneity, some meta-analyses have narrowed inclusion criteria (eg, one review analysed only studies providing HRs, comprising 20 studies involving 91 391 296 people and reported an HR of 1·03 (95% CI 1·02–1·05) per 1 μg/m3 increment in PM2·5. A conservative pooled estimate, obtained from a meta-analysis of five studies that used active case ascertainment of high-quality studies, reported an HR of 1·17 (0·96–1·43) per 2 μg/m3 increment in PM2·5, although CIs were wide and included the null. Pooled HRs were reported from five studies each of nitrogen dioxide (1·02 [0·98–1·06] per 10 μg/m3) and nitrogen oxides (1·05 [0·98–1·13] per 10 μg/m3) and four studies of ozone (1·00 [0·98–1·05] per 5 μg/m3), none of which were significant. Other pollutants have been assessed by too few studies for meta-analysis.
In both HICs and LMICs, where air pollution is often high and increasing, PM2·5 and PM10 concentrations have been associated with dementia, MCI, and Alzheimer’s disease. Ambient (ie, outdoor) and household (ie, indoor) air pollution might have distinct or synergistic risks. Studies in LMICs have shown that compared with clean fuel, solid fuel use, a proxy for household air pollution, is associated with higher dementia risk and accelerated cognitive decline among middle-aged and older adults (ie, aged >45–50 years). Residential wood and coal burning stoves are a source of indoor air pollution, and are reported to contribute 38% of the UK’s PM2·5 emissions and associated health risks.
A US, 7-year, cohort study of more than 18 million participants reported that the PM2·5 constituent with the strongest association with dementia risk was black carbon (HR 1·12 [95% CI 1·11–1·14] per 1 μg/m3 increment). The studies have mainly been in older adults at baseline, but this factor does not rule out an effect earlier in life.
A longitudinal study with a mean follow-up of 6 years in 2927 Swedish residents (1845 [63%] women and 1082 [37%] men who did not have dementia at baseline; baseline mean age of 74 years [SD 10·7]) considered PM2·5 and nitrogen oxide yearly from 1990 to examine whether cardiovascular disease (ie, atrial fibrillation, ischaemic heart disease, heart failure, and stroke) modified or mediated the association between pollution and dementia and reported that it did. The effect of air pollution is worst among people with these pre-existing conditions.
There is emerging evidence on the potential effects of improved air quality on cognitive decline and dementia incidence. A French cohort study with a 12-year follow-up reported that a 12·2 μg/m3 reduction in median PM2·5 between 1990 and 2000 was associated with a decreased risk of dementia (HR 0·85, 95% CI 0·76–0·95). Larger air quality improvement (reduction in PM2·5 and NO2 over 10 years) was associated with lower dementia risk in older US women. In a quasi-experimental study, China’s Air Pollution Prevention and Control Action Plan mitigated cognitive decline in older adults, indicating that strict clean air policies might reduce the risk of cognitive ageing (measured by the MMSE) associated with air pollution. A difference in China’s central heating policies between the north and the south led to differences in air pollution concentrations, and higher air pollution (ie, PM10, NO2, SO2, CO, and O3) in the northern sample was associated with a 42·4% higher dementia risk than for the southern sample.
As the evidence base grows, it would be valuable to standardise study design, reporting, and analyses to allow comparisons and achieve a granular understanding of the association between air pollution and dementia. Given the close link between socioeconomic circumstances, household conditions, and exposure to air pollution, minimising residual confounding in these studies is difficult.
Overall, there is increasing support for the implementation of WHO global air quality guidelines that ultimately aim for mean annual PM2·5 concentrations of less than 5 μg/m3. It is unclear whether any safe concentration of air pollution exists, as every 1 μg/m3 increase in PM2·5 is associated with increased dementia risk. The lowest annual PM2·5 concentration in global megacities was 6·7 μg/m3 in Miami, FL, USA, and the top five most polluted cities had annual mean concentrations of PM2·5 between 89 μg/m3 and 149 μg/m3. Little is known about risk in relation to dementia subtypes and whether individual particulate matter constituents are important (eg, black carbon, sulphates, nitrates, and ammonium).
[Untreated vision loss]
[..] Our Commission has not previously considered vision loss as a risk factor for dementia, but considerable new evidence has emerged. This evidence includes a meta-analysis of 14 prospective cohort studies, with follow-up of 3·7–14·5 years, including 6 204 827 older adults who were cognitively intact at baseline, of whom 171 888 developed dementia. Vision loss was associated with a pooled RR for dementia of 1·47 (95% CI 1·36–1·60; figure 8). In an accompanying meta-analysis of 12 prospective cohort studies with 45 313 participants, 13 350 developed cognitive impairment, and the RR for vision loss and future cognitive impairment was 1·35 (1·28–1·41).
A second meta-analysis identified an increased risk of all-cause dementia (RR 1·38, 95% CI 1·19–1·59; n=37 705) with visual loss. When broken down into different eye conditions, an increased dementia risk was associated with cataracts (three studies; 6659 participants; 1312 cases; HR 1·17, 95% CI 1·00–1·38; I2=0·0%) and diabetic retinopathy (four studies; 43 658 participants; 7060 cases; 1·34, 1·11–1·61; I2=63·9%), but not with glaucoma (six studies; 175 357 participants; 44 144 cases; 0·97, 0·90–1·04; I2=51·5%) or age-related macular degeneration (three studies; 7 800 692 participants; >2559 cases, exact number could not be determined; 1·15, 0·88–1·50; I2=91·0%). [..]
A US study followed up 3038 older adults (ie, aged >65 years) with cataracts and healthy cognition at baseline for more than 20 years. The analysis controlled for age, race, APOE genotype, education, smoking, and an extensive list of comorbidities and reported that people who had cataract extraction had significantly reduced dementia risk compared with those who did not have cataract extraction (HR 0·71, 95% CI 0·62–0·83; 23 554 person-years of follow-up). Although a UK Biobank study of 300 823 people reported that people with cataracts had an increased risk of dementia (1·21, 1·01–1·46), there was no difference in dementia risk between those who had cataract surgery and healthy controls.
[The researchers went on to assign a percentage of each risk factor that could contribute to dementia. The top five risk factors were: hearing loss (midlife [18-65 years], 7%), high LDL cholesterol (midlife 7%), less education (early life [<18 years], 5%) and social isolation (late life [>65 years], 5%). Depression (midlife), traumatic brain injury (midlife) and air pollution (late life) each contributed three percent.]
[..] The advent of disease-modifying drugs for dementia is a long-awaited scientific breakthrough, but results vary from modestly positive to neutral, and the clinical implications are still unclear. There is exciting progress in the field of biomarkers, but biomarkers are not enough by themselves to justify diagnosis. Clinically, biomarkers should be used only to help to classify the neuropathology in people with dementia, particularly that of Alzheimer’s disease. Drug and psychosocial treatments are progressing, and there are more people living with dementia than ever before. It is even more important now, therefore, that care for people with dementia and their families is improved.”
Full report, G Livingston, J Huntley, KY Liu et al. The Lancet, 2024.7.31