Table e-1. Summary data for cross-sectional, case-control and cohort study designs.
First author, yearCountry
Total N / Setting
(study name) / Groups / Age (y)
Mean ± SD
(min-max) / Female / Vitamin D / Cognitive Test or
Dementia Criteria / Results2
Method1 / Cut point
(nmol/L)
Cross-sectional studies
Annweiler, 2009(30)France
N=7,598 / Community
EPIDOS / 752 randomized subset of participants / >75 / 100% / RIA
DiaSorin / < 25
25
25-50
50 / SPMSQ / Lower SPMSQ score in lower compared to higher vitamin D group but no linear association. Significant difference in SPMSQ score between all vitamin D groups.Odds ratio for low vitamin D and SPMSQ (<8) was significant in the crude and fully adjusted model.
AF: age, BMI, calcium, chronic diseases, depression, education, hypertension, PTH, physical activity, psychoactive drugs
Aung, 2006(25)
USA
N=139** / Community
CREST / 70 self-neglecting elders
69 controls / >65 years / 65.7%
62.3% / RIA
DiaSorin
/ < 2525-50
50 / Clock drawing test
MMSE / No significant difference between MMSE or clock drawing test between any vitamin D group.
Benge, 2009(34)
USA
N=111 / Hospital / 39 PHPT with CSI3
72 PHPT without CSI / 60.5±12.3 / 82.9% / Unknown / COWA
LGPT
HVLT-R
PASAT (2 and 3 second)
Stroop test (1, 2, 3)
Trails A and B
WAIS-III digit span, WAIS-III digit symbol / No significant difference between the group with CSI and group without CSI.
Buell, 2009(35)
USA
N=1253 / Community / 1080 participants / 75.0±8.5 / 75.9% / RIA
DiaSorin / < 25
25-50
50 / Block design
COWA test
Digit span
Digital symbol coding
Matrix reasoning
Trails B
WMS-III recognition
WMS-III LM recognition / Significant for all tests for at least one pairwise comparison.
AF: age, alcohol, BMI, center education, kidney function, physical activity, race, season, sex
Multiple regression data in table e-2
Buell, 2010(36)
USA
N=318 / Community
Homecare agencies / 318 participants / 73.5±8.1 / 72.6% / RIA
DiaSorin / NINCDS-ADRDA (AD)
NINDS-AIREN (VTD)
DSM-IV (other dementia) / Significant difference between AD or all-cause dementia and no stroke or dementia.
All-cause dementia diagnosis:
Beta coefficient was significant in the univariate and final models. Odd ratio in the final model was 2.21 (1.13, 4.32).
Alzheimer’s disease diagnosis:
Beta coefficient was significant in the univariate and final models. Odd ratio in the final model was 2.65 (0.99, 7.16).
AF: age, sex, race, BMI, education, kidney function, multivitamin use, season, diabetes, homocysteine, hypertension, ApoE.
El-Ghoneimy 2009(37)
Egypt
N=30 / Unknown / 30 participants with multiple sclerosis / 28.9 / 63.3% / ELISA
Bioscientia / < 50 / Faces Symbol Test
PASAT / Correlation between test score and vitamin D not significant.
Hii, 2004(31)
Australia
N=107 / 107 participants with low trauma fracture admitted to an aged care centre / 83 (median) / 81.3% / Unknown / MMSE / No association between vitamin D and MMSE score.
Houston, 2007(38)
Italy
N=1155 / Community
InCHIANTI / 435 men
541 women / 74.8 for combined groups / 0%
100% / RIA
DiaSorin / < 25
25-50
50 / MMSE / There was a significant difference in MMSE score between groups for women only.
AF: age, sex
Lee, 2009(39)Italy, Belgium, Poland, Sweden,
UK, Spain, Hungary, Estonia
N=3,369 / Community
EMAS / 3133 participants with vitamin D and cognitive test data available / 59.9±11.0 / 0% / RIA
DiaSorin / <25
<50
50-75
75 / CTRM test
DSST
ROCF copy
ROCF recall / Only the DSST was significant when compared to vitamin D (per 10 mmol/L or by group ( 75 as reference).
AF: age, alcohol, BMI, center depression, education, physical activity, physical performance, season, smoking
Llewellyn, 2009(28)
UK
N= 1,766 / Community and institutionalized
Health Survey of England / 212 cognitively impaired
1,554 cognitively normal / 83.3±7.9
77.6±8.5 / 67.9%
58.8% / RIA
DiaSorin / Q1: ≤ 30
Q2: 44
Q3: 65
Q4: 66 / AMT / Quartile group differences were significant between impaired and normal groups. Odds ratio between AMT score (<8) and vitamin D (75 nmol/ as reference) was significant in unadjusted and adjusted models.
AF: age, albumin, alcohol, education, ethnicity, impaired mobility, morbidity (diabetes, heart disease, hypertension, psychiatric), season, sex, smoking
Liu, 1997(27)
Canada
N=155* / Institutionalized / 155 participants in a long-term care home / 83.2±7.1 / 50.3% / CPBA
In-house / < 25
25-50
50 / SPMQ / Odds ratio for low vitamin D and SPMQ ≤ 6 not significant by logistic regression. No difference in SPMQ scores within or between vitamin D groups in March or Sept.
McGrath, 2007(40)
USA
N=17,099 / Community
NHANES III / 4,747 younger adults
4,809 older adults
Subset of participants with both vitamin D and cognitive test data4 / (20-59)
(60-90) / N/A / RIA
DiaSorin
/ Q1: <39.5Q2: 52.9
Q3: 66.9
Q4: 84.9
Q5: >84.9 / Mean reaction time
Memory and learning
Symbol-digit substitution
Serial digit learning / Between quintile groups not significant for any cognitive test except for the memory and learning score in the elderly group.
AF: activity, age, race/ethnicity, sex
Ogihara, 1990(41)
Japan
N=60 /
Hospital
/ 22 SDAT cases20 VTD cases
18 non-demented cases / 81±7
80±8
78±6 / 100%
100%
100% / CPBA
In-house / Dementia Screening Scale of Hasegawa score
Further classified according to Ischemic score
(SDAT, VTD) / No significant difference between any group comparisons.
Oudshoorn, 2008(24)
Netherlands
N=225* /
Clinic
/ 225 AD patients from an geriatric outpatient clinic / 77.6±7.3 / 65% / RIADiaSorin
/ < 50 50
< 25
25-50 / MMSE / Significant difference in MMSE between
lowest and highest vitamin D groups.
Beta coefficient was significant in the crude and
adjusted models.
AF: action radius, age, B1, B6, B12, education, mobility score, sex
Perez-Llamas, 2008(26)
Spain
N=86* / Institutionalized / 86 participants in public nursing homes / 77.4±8.1 / 66.3% / HPLC
CIC
/ < 50 50
< 25
25-50 / SPMSQ / No difference in SPMSQ between vitamin D groups and no correlation between cognitive score and vitamin D. Significant difference in SPMSQ between the lowest and middle vitamin D groups.
Przybelski, 2007(29)
USA
N=80* / Clinic / 80 patients with cognitive symptoms from an outpatient clinic / 79.5
±1.6 (SE) / N/A / RIA
DiaSorin
/ < 2525-50
50 / MMSE / Correlation between vitamin D and MMSE significant. Significant difference in MMSE between the lowest and middle vitamin D groups.
Sakuma, 2006(42)
Japan
N=103 / Hospital / 50 acute hip fracture cases / 82.6±8.7 / 82% / ELISA
DiaSorin / Dementia severity criteria (DI-DIV) of the long-term care insurance system developed by the Japanese Ministry of Health, Labour and Welfare, 1993 / No significant difference between any dementia group (DI, DII, DIII-DIV) compared to cognitively normal group.
Wilkins, 2009(23)
USA
N=60 / Community / 30 African Americans
30 European Americans
Matched on CDR score
/ 75±8.2 / N/A / RIADiaSorin / < 50
50 / MMSE
Short blessed test / All participants with lower compared to higher vitamin D had poorer performance on the SBT, but for the MMSE significance was seen only in the African American group. Correlation between MMSE in low vitamin D group.
AF: age, education, race
Wilkins, 2006(43)USA
N=80 / Community / 40 mild dementia
40 non-demented / 74.8±7.7 / 62.5% / RIA
DiaSorin
/ < 2525-50
50 / CDR sum of boxes
Factor score
MMSE
Short blessed test / Significant difference in CDR sum or boxes score and short blessed test score between lowest and highest vitamin D groups.
AF: age, race/ethnicity, season, sex
Woo, 1991(44)
China
N=80 / Institutionalized / 80 participants in an old age hostel / 76 / 100% / Unknown / Mental test score / No significant difference in vitamin D concentration between MTS score groups of <4 and ≥4
Case-control studies
Evatt, 2008(46)USA
N=296 / CRIN registry database / 97 AD cases
99 controls
Matched on age, sex, race/ethnicity, APOE, residence / 66.4 (47-88)
65.7 (39-89) / 43%
43% / ELISA
IDS / NINCDS-ADRDA (AD) / No significant difference between cases and controls.
Ferrier, 1990(47)
UK
N=61 / Hospital (admitted, clinic)
Community / 26 ATD cases
24 controls
Matched on age, sex / 76±7
74±5 / 76.9%
75% / RIA
In-house / DSM-III
MTS score <30
HII score <7
(ATD) / Significant difference between cases and controls.
Jorde, 2006(48)
Norway
N=202 / Community
Tromsø / 148 non-smokers / 62.1±13.6 / 45.9% / RIA
DiaSorin / CalCAP
COWA
Digit span forward
Digit symbol
Seashore rhythm
Stroop test (1, 2 and 3)
Trails A and B
Verbal Recall
Visual Recall
WAIS Vocabulary
Word List test / All beta-coefficient t-values were not significant by multiple linear regression.
Kipen, 1995(49)
Australia
N=60 / Outpatient clinic
Twin volunteer registry / 20 mild dementia cases
40 controls
Matched on age
/ 76.4±4.572.1±6.7 / 100%
100% / CPBA
Buhlman / DSM-III-R
NINCDS-ADRDA
(Mild dementia) / Significant difference between cases and controls.
Luckhaus, 2009(50)
Germany
N=47 / Outpatient clinic / 20 AD cases
19 MCI cases
8 controls
Matched on age / 70.4± 8.2
67.3±10.9
72.4±6.1 / 40%
47%
50% / ELISA
IDS / NINCDS-ADRDA (AD)
Petersen, 1999 (MCI) / Significant difference between all group comparisons.
Martyn, 1989(51)
UK
N=61 / Hospital / 27 AD cases
34 controls / 78.5 (58-90) 78.4 (66-89) / 100%
100% / RIA
Unknown / Intellectual deterioration over a period of more than 6 months and no other indication of dementia / Significant difference between cases and controls.
Nes, 1988(52)
Norway
N=32 / Community
(cross-sectional socio-medical) / 16 dementia cases
16 controls
Matched on age, sex / 82 (76-87) 82 (76-87) / 56%
56% / HPLC-UV
In-house / DSM-III / Significant difference between cases and controls and also in subgroups not taking vitamin D supplements.
Sato, 2005(53)
Japan
N=200 / Clinic
Community / 58 severe AD cases
42 mild AD cases
100 controls
Matched on age
/ 79.7±4.579.9±5.2
80.6±6.8 / 100%
100%
100% / CPBA
Nichols / DSM-III-R
MMSE / Significant difference between all group comparisons.
Sato, 1998(54)
Japan
N=186 / Institutionalized
Community / 46 AD cases
140 controls
Matched on age
/ 81.3±5.480±4.7 / 100%
100% / CPBA
Nichols / DSM-III-R
NINDS-ADRDA / Significant difference between cases and controls.
Walker, 2009(55)
USA
N=128 / Hospital
Community / 39 patients with PHPT who underwent surgery
89 controls / 61.3±1.0
55.6±0.4 / 100%
100% / RIA
DiaSorin / Book category
NAART
Rosen target detection
RVDLT
Serial digit learning
WAIS-R digit symbol
WAIS digit span
WMS logical memory / No linear association between vitamin D at baseline and any abnormal cognitive test score. No relationship between change in cognitive test and change in vitamin D before and after surgery.
Cohort studies
Llewellyn, 2010(57)Italy
N=858
Mean follow-up:
2 years / Community
InCHIANTI / 858 participants / 74.0±6.8 / 56.8% / RIA
DiaSorin / Q1: <25
Q2: <50
Q3: <75
Q4: >75 / MMSE
Trails A and B / Multivariate adjusted relative risk of lower MMSE (decline of 3 or more points) and Trials B (the worst 10% or test discontinue) but not Trials A in the low (Q1) compared to high (Q4) vitamin D groups was significant. The trend was also significant.
AF: Alcohol, age, BMI, cognitive score (baseline), depressive symptoms, education, energy intake (total), impaired mobility, season, sex, smoking, vitamin E
Slinin, 2009(32)
USA
N=1,606
Mean follow-up:
4.6 years / Community
MrOS / 1,604 (3MS result)
1,564 (Trails B) / 65 years / 0% / LC-MS
In-house / Q1: ≤49.8
Q2: 62.6
Q3: 74.4
Q4: ≥74.4 / 3MS
Trails B / Multivariate adjusted odds ratio was not significant for prevalent or incident cognitive impairment (3MS <80 at baseline or decline of 5 points or more on follow-up; Trails B time <1.5 SD above mean).
AF: Alcohol, age, BMI, education, health (self-reported), IADL, physical activity, race/ethnicity, season, site, smoking
1 Method type and source. In-house refers to a non-commercial method.
2 Details of the statistical analyses can be found in the table e-2 (p values, beta coefficients, odds ratios, adjustment factors, test type).
3 CSI based on having 3 or more of 15 cognitive tests below -1.5.
4The mean reaction time score (n=4,747), symbol-digit substitution coding speed (n=4,688), and serial digit learning trials to criterion (n=4,584) was done only in the younger adult group whereas the memory and learning score was performed only in the older adult group.
* Additional study data obtained through author correspondence. This new information appears in italics.
** Additional participants recruited since publication; additional data obtained through author correspondence.
AD, Alzheimer disease; AF, adjustment factors; AMT, abbreviated mental test; ATD, Alzheimer type dementia; CalCAP, California Computerized Assessment Package; CDR, Clinical Dementia Rating Scale; CIC, Chromsystem Insturments & Chemicals GmbH; COWA, controlled oral word association;CRIN, Clinical Research in Neurology database; CSI, clinically significant impairment; CTRM, camden topographical recognition memory test; DSM (-III/-III-R/-IV), Diagnostic and Statistical Manual of Mental Disorders (editions); DSST, digit-symbol substitution test; ELISA, enzyme linked immunosorbent assay; EPIDOS, Epidémiologie de l’Ostéoporose; EMAS, European Male Aging Study; HPLC, high-performance liquid chromatography; HII, Hachinski Ischaemic Index;HVLT-R, Hopkins verbal learning test - revised (total recall, delayed recall, delayed recognition); IADL, instrumental activities of daily living; LGPT, Lafayette grooved pegboard test (dominant, non-dominant hand); LCMS, liquid chromatography mass spectroscopy; MAE, multilingual aphasia examination; MCI, mild cognitive impairment; MMSE: mini-mental status examination; 3MS, modified mini-mental state examination; MTS, Mental Test score; NAART, North American adult reading test; NAME, Nutrition and Memory in Elders study; NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer's Disease and Related Disorders Association; NINDS-AIREN, National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherché et l’Enseignement en Neurosciences; PASAT, paced auditory serial addition task; PHPT, primary hyperparathyroidism; Q, quartile or quantile; RIA, radioimmunoassay; ROCF, Rey-Osterrieth complex figure test; RVDLT, Rey visual design learning test; SDAT, senile dementia of the Alzheimer type; SPMSQ: short portable mental status questionnaire; WAIS, Wechsler adult intelligence scale; WMS, Wechsler memory scale; VTD, vascular type dementia.
Table e-1 Page 1