Table 2 Results of included studies

Afzal et al., 2014 [1], prospective cohort study

E1: 3715

E2: 4087

E3: 2384

Incidence of AD or dementia (ICD 8th and 10th edition diagnoses entered in the national Danish Patient Registry and the national

Danish Causes of Death Registry)

Gender, age, smoking status, BMI, leisure time and work-related physical activity, income level, education, diabetes mellitus, hypertension, alcohol consumption, cholesterol, creatinine, month of blood sample, seasonal adjusted vit D concentrations

418 subjects developed AD and 92 subjects vascular dementia, 14 subjects had both diagnoses.

Risk of developing dementia:

Analysis adjusted for all measured confounders:

AD: E1 = reference,

E2: HR = 1.23 (95% CI 0.97–1.55),

E3: HR = 1.29 (95% CI 1.01–1.66) (p = 0.03)

Vascular dementia: E1 = reference,

E2(<50 nmol/L or 20 ng/mL): HR = 1.22 (95% CI 0.79–1.87),

(p = 0.42)

Combined: E1 = reference,

E2: HR = 1.24 (95% CI 1.00–1.54),

E3: HR = 1.27 (95% CI 1.01–1.60) (p = 0.02)

unclear

Annweiler et al., 2011 [2], prospective cohort study

E1: 33 (subtle cognitive impairment [2])

E2: 7 (subtle cognitive impairment [1])

Incidence of dementia (diagnosed by experts, according to DSM IV, NINCDS-ADRDA)

Subtle cognitive impairment at baseline, presence of cardiovascular risk factors at baseline (age >85 years, hypertension, diabetes mellitus, BMI >25, lack of physical activity, smoking), diagnosis of Parkinson’s disease at baseline

10 women developed dementia, 4 of these AD

E1: 3

E2: 7 (4 AD)

Risk of developing Non-Alzheimer’s dementia

Analysis adjusted for all measured confounders:

E1 = reference

E2: OR = 19.57 (95%CI 1.11–343.69)

Risk of developing AD

Unadjusted analysis:

E1 = reference

E2: OR = 1.06 (95%CI 0.97–1.15).

high

Graf et al. 2014 [3], prospective cohort study

E1: 15 (cognitively normal [11], MCI [4])

E2: 33 (cognitively normal [27], MCI [6]

E2: 58 (cognitively normal [52]; MCI [6])

E3: 140 (cognitively normal [110, MCI [30])

Incidence of dementia (diagnosed by experts, validated cognitive scales, DSM IV-TR, NINCDS-ADRDA, ADDTC, and NINDS-AIREN)

Gender, age, education level, basic (BADL) and instrumental (IADL) activities of daily living, comorbidities (CIRS), calcaemia, Vit B12 status, ApoE Eε4 genotype, mini nutritional assessment, albuminaemia, BMI

46 subjects developed dementia, 28 cognitively normal subjects and 18 with MCI.

Analysis adjusted for all measured confounders:

E1 = reference

E2: RR = 2.87 (95% CI 0,36–22,77)

E3: RR = 6.18 (95% CI 0,87–43,76)

E4: RR = 2.85 (95% CI 0,45–17,95)

Reclassification for Meta-Analysis:

Analysis adjusted for all measured confounders:

e1 = reference

e2: RR = 4.55 (95% CI 1.04–19.82)

e3: RR = 1.35 (95% CI 0.39–4.61)

unclear

Knekt et al., 2014 [4], retrospective cohort study

E1: 1240

E2: 1258

E3: 1216

E4: 1296

Incidence of dementia leading to hospitalisation (ICD 8 from the nationwide Finnish hospital discharge register or death certificates from Statistics Finland)

Gender, age, month of blood sample, education level, marital status, leisure time physical activity, smoking status, BMI, alcohol consumption, hypertension, plasma fasting glucose concentration, serum triglyceride concentration, serum total cholesterol concentration

151 subjects developed dementia, 34 men and 117 women.

E1: 21 (m = 13, f = 8)

E2: 33 (m = 12, f = 21)

E3: 37 (m = 5, f = 32)

E4: 60 (m = 13, f = 47)

Analysis adjusted for all measured confounders:

Men: E1: HR = 0.74 (95% CI 0.29–1.88)

E2: HR = 0.63 (95% CI 0.25–1.56),

E3: HR = 0.41 (95% CI 0.14–1.19),

E4 = reference

Women: E1: HR = 0.33 (95% CI 0.15–0.73)

E2: HR = 0.60 (95% CI 0.34–1.06),

E3: HR = 0.90 (95% CI 0.56–1.44),

E4 = reference

Combined: E1: HR = 0.48 (95% CI 0.28–0.84)

E2: HR = 0.62 (95% CI 0.39-1.00),

E3: HR = 0.75 (95% CI 0.49-1.14),

E4 = reference

unclear

Littlejohns et al., 2014 [5], prospective cohort study

E1: 1169

E2: 419

E3: 70

Incidence of dementia (diagnosed by experts, annual cognitive assessments, NINCDS-ADRDA)

Age, season of vit D collection, education, gender, BMI, smoking, alcohol consumption, depressive symptoms, diabetes, hypertension, ethnicity, income, occupation

171 subjects developed dementia, 102 of these AD

E1: n.s.

E2: n.s.

E3: n.s.

Analysis adjusted for age, season of vit D collection, education, gender, BMI, smoking, alcohol consumption, depressive symptoms:

Dementia: E1 = reference,

E2: HR = 1.53 (95% CI 1.06–2.21),

E3: HR = 2.25 (95% CI 1.23–4.13) (p = 0.002)

AD: E1 = reference,

E2: HR = 1.69 (95% CI 1.06–2.69),

E3: HR = 2.22 (95% CI 1.02–4.83) (p = 0.008)

Similar results for analysis that additionally adjusted for diabetes and hypertension (data not shown).

unclear

Schneider et al. 2014 [6], prospective cohort study

E1: Whites 285; Blacks 267

E2: Whites 283; Blacks 272

E3: Whites 284; Blacks 261

Incidence of AD or dementia leading to first hospitalisation (ICD 9 from hospital discharge records)

Gender, age, education, income, smoking, alcohol consumption, physical activity, BMI, waist circumference, use of vit D supplements, diabetes, hypertension, use of hypertension medication, cholesterol, estimated glomerular filtration rate, calcium status, phosphate, PTH, season adjusted vit D concentrations

145 subjects developed AD or dementia.

E1: Whites 18; Blacks 23

E2: Whites 31; Blacks 24

E3: Whites 24; Blacks 25

Analysis adjusted for age, gender, education, income, physical activity, smoking, alcohol consumption, BMI, wait circumference, use of vit D supplements:

Whites: E1 = reference,

E2: HR = 1.74 (95% CI 0.95–3.18),

E3: HR = 1.32 (95% CI 0.69–2.55)

Blacks: E1 = reference,

E2: HR = 1.22 (95% CI 0.68–2.19),

E3: HR = 1.53 (95% CI 0.84–2.79)

Reclassification for Meta-Analysis:

145 subjects developed AD or dementia.

e1: 75 out of 876

e2: 63 out of 694

e3: 7 out of 82

Analysis adjusted for age, gender, education, income, physical activity, smoking, alcohol consumption, BMI, waist circumference, use of vit D supplements:

e1 = reference

e2: HR = 1.22 (95% CI 0.85-1.74)

e3: HR = 1.44 (95% CI 0.65-3.21)

unclear