Skip to main content

Table 6 Individual studies reporting the association between other trace elements and dementia

From: Environmental risk factors for dementia: a systematic review



Sample description





Still & Kelley, 1980 [51]


All first admissions of people aged ≥55 years to South Carolina Department of Mental Health hospitals from three counties who had lived in that county for at least 10 years.

Case records for 160 patients were examined and, based on that evidence, 67 were diagnosed with a primary degenerative dementia, i.e. AD.

Cross-sectional study. Annual incidence of dementia was calculated per county, based on the admissions data. These were then compared with fluoride concentrations in public water supplies (time of measurement is not given).

Horry County was reported to have highest levels of fluoride (4.18 ± 0.19 mg/l) compared to Anderson (0.49 ± 0.10 mg/l) and York (0.61 ± 0.12 mg/l) Counties. It also has the lowest calculated annual incidence of dementia per 100,000 population (3.6; Anderson 20.8, York 17.1).


Loef, Schrauzer, & Walach, 2011 [23]


Systematic review of studies relating selenium to AD from 8 databases.

56 studies: 9 placebo-controlled, 4 prospective, 4 cross-sectional, 15 case–control, and 24 autopsy studies.

Systematic review.

One double-blind RCT is reported (PREADVISE[5254]) but this seems to have subsequently converted into an observational study.


A prospective cohort study is referred to by citing a conference abstract [55]; subsequent reports do not focus on selenium.[56]

Of 15 case–control studies, the authors comment that “four studies reported about increased levels of Se-concentration or GPx-activity while the majority found decreased levels, albeit non-significant in some studies.” (p. 87).

Of the 24 autopsy studies, the authors conclude that “the autopsy studies do not yield a consistent picture of whether, how and where in brain Se levels become altered in subjects with AD.” (p. 87).

Gillette-Guyonnet et al., 2005 [57]


Toulouse subset of the EPIDOS study cohort of women aged ≥75 years.

1462 women from one centre of whom 60 developed clinically diagnosed AD. 323 had normal cognitive function and the remainder were lost to follow up.

Prospective cohort study: logistic regression. Water consumption based on self-report (at baseline and follow up) was combined with local tap water composition data. Dementia was clinically diagnosed.

Low silica consumption in water (≤4 mg vs >12 mg per day) was associated with an increased risk of dementia (multivariable-adjusted OR, 95 % CI: 2.74, 1.09-6.86). Women who developed AD had lower silica intake at baseline and showed a decrease in daily silica intake at follow up.


Rondeau et al., 2009 [39]


PAQUID: A community-based cohort of 3,777 elderly people aged ≥65 years in SW France.

1925 individuals of whom 461 developed clinically diagnosed dementia (364 AD).

Prospective cohort study: Cox PH models. Mean levels of silica in drinking water over the previous decade based on current residential location was linked to incident dementia over 15 years follow up.

Highest:lowest quartile of silica in drinking water was associated with an increased risk of dementia and AD (multivariable-adjusted HR, 95 % CI: 1.33, 1.01-1.74; 1.33, 0.98-1.80).


Taylor et al., 1995 [48]


Same cohort as Forster et al., 1995 [47]

Water samples were obtained for 214 addresses of the 218 cases and controls.

Cross-sectional study (case–control). Aluminium concentration in water samples drawn from the place of residence at which they had lived longest within 10 years prior to onset of dementia (or equivalent date for controls).

There were no differences in silica concentrations in samples for cases or controls and no association between increased levels of silica and dementia (OR, 95%CI ≥3 mg/L 0.8, 0.34-1.83).


Martyn et al., 1997 [46]


Participants were selected from CT records of eight neuroradiology centres in the UK.

A total of 872 men (106 AD, 99 other dementia, 226 brain cancer, 441 other disease of the nervous system)

Cross-sectional study (case–control): logistic regression. Average levels of silicon in drinking water based on residential history (after age 25 years) was related to diagnosis based on hospital records.

There were no associations identified between silicon concentrations consistently above 6 mg/L (compared to lower levels) with each of the three comparison groups (e.g. AD vs other diagnoses (adjusted-OR, 95 % CI: 0.94, 0.39-2.26).


  1. AD Alzheimer’s dementia, CI confidence interval, CT computed tomography, HR hazard ratio, OR odds ratio, PH proportional hazards, RCT randomized, controlled trial, RR relative risk, SD standard deviation, SMR Standardised Mortality Ratio, UK United Kingdom