Table E-4. Inclusion, Exclusion Criteria and Details of the Statistical Analyses for All

Table e-4. Inclusion, exclusion criteria and details of the statistical analyses for all 37 studies.

First Author,
year / Inclusion, Exclusion Criteria / Cognitive Test or
Dementia Criteria / Vitamin D (nmol/L) or Dementia Group: Unadjusted Mean (SD)1 of Cognitive Test Score or Serum 25OHD1 Concentration, P-value, Statistical Test

Cross-sectional studies

Annweiler, 2009(30) / Inclusion: randomized subset of the EPIDOS study, community-dwelling women age ≥75 from 1 of 5 cities in France chosen from electoral lists.
Exclusion: inability to walk independently, hip fracture or bilateral hip replacement, inability to understand or answer the study questionnaires. / SPMSQ
Cognitive impairment:
(SPMSQ <8) / <25 (n=129): 8.56 (1.67); ≥25 (n=623): 9.05 (1.34), p <0.001, t-test.
25-50 (n=393): 9.12 (0.06)*
≥50 (n=230): 8.94 (0.11)*
Beta coefficient (95%CI):
-0.003 (-0.012, 0.006), p=0.512, linear association
Adjusted OR (95%CI) of cognitive impairment with 25OHD concentration <25:
Crude: 2.08 (1.22, 3.55); p=0.007; Full: 1.99 (1.13, 3.52); p=0.017, logistic regression.
Stepwise backward: 2.03 (1.17, 3.53); p=0.012, adjusted logistic regression.
Aung, 2006(25) / Inclusion: age ≥65; validation by Adult Protective Services as elder self-neglect (SN) from the CREST pilot study; resident of Harris County, Texas; English-speaking.
Exclusion: Validated cases of elder-self-neglect for whom blood samples were not available. / MMSE
Clock drawing test
(% Abnormal) / ≥50: SN (n=23): 24.48 (4.57)**; Controls (n=27): 25.55 (4.15)**
25-50: SN (n=25): 24.56 (3.58)**; Controls (n=29): 24.17 (4.33)**
<25: SN (n=17): 23.35 (3.9)**; Controls (n=7): 26.57 (2.51)**
Percent, SN only:
<25 (n=13): 31%, ≥25 (n=28): 21%, p=0.517, χ2 test.
<50 (n=31): 26%, ≥50 (n=10): 10%, p=0.223, χ2 test.
Benge, 2009(34) / Inclusion: patients diagnosed with PHPT and considered for parathyroidectomy and referred for neuropsychological evaluation.
Exclusion: pre-surgical neurological conditions (most commonly due to history of head injury with loss of consciousness or long standing history of dementia); long time spans between baseline evaluation and date of surgery; lack of baseline evaluation; diagnosed multiple endocrine neoplasias; previous parathyroidectomy; long-standing psychiatric histories; required a translator for testing. / CSI2 / No significant differences on baseline biochemical parameters were found between CSI (n=39) and no CSI (n=72), p-value not reported, t-test.
Buell, 2009(35) / Inclusion: using one of four Boston homecare agencies (ASAPs; criteria for receiving services include: age ≥60, low income, diminishment in activities of daily living, and an unmet need in a critical area such as food or personal care); English speaking, free of severe auditory and visual impairment and specific conditions (epilepsy, human immunodeficiency virus, schizophrenia, bipolar disorder, mental retardation, brain tumour).
Exclusion:MMSE score ≤10 or verbal IQ (estimated with NAART) <75; do not speak English, have severe visual or auditory impairment, are illiterate, have contact prevented by caregiver or refuses referral, or have a history or current diagnosis of epilepsy, HIV/AIDS, schizophrenia, bipolar disorder, mental retardation, or brain tumour documented in the ASAP's case records. / MMSE
Trails A (seconds)
Trails B (seconds)
DSST
Matrix reasoning
Block design
WMS-III recognition
WMS-III LM recognition
COWA
Digit span / Adjusted mean (SD):
<25 (n=192): 25.3; 25-50 (n=512): 24.1; ≥50 (n=376): 26.0; p<0.05†
<25: 92; 25-50: 87; ≥50: 82.4; p<0.05†.
<25: 224; 25-50: 208; ≥50: 205; p<0.05†.
<25: 32.6; 25-50: 33; ≥50: 36; p<0.05†.
<25: 7.2; 25-50: 7.0; ≥50: 7.7; p<0.05†.
<25: 18.6; 25-50: 20; ≥50: 21; p<0.05†.
<25: 20.5; 25-50: 20.3; ≥50: 20.3; p>0.05
<25: 22.9; 25-50: 22.3; ≥50: 22.8; p>0.05
<25: 27; 25-50: 27.6; ≥50: 28.3; p>0.05
<25: 13.1; 25-50: 13.8; ≥50: 14.2; p<0.05†, ANCOVA.
Beta coefficient (SE), partial correlation value:
Executive function factor (n=931): 0.01 (0.003), 0.13; p=0.001
Memory function factor (n=931): -0.001 (0.004), -0.002; p=0.65
Attention/processing speed factor (n=931): 0.01 (0.003), 0.08; p=0.03, multiple regression. AF: age, sex, race, education, BMI, kidney function, center, season, physical activity, alcohol use, homocysteine, ApoE allele status, plasma B vitamins, multi-vitamin use (y/n).
Buell, 2010(36) / Inclusion: free of severe auditory and visual impairment, schizophrenia, bipolar disorder, mental retardation.
Exclusion: MMSE score ≤10, or verbal IQ (estimated with NAART) <75. / NINCDS-ADRDA (AD)
NINDS-AIREN (VTD)
DSM-IV (other) / AD (n=41): 42.2 (15.7)
All-cause dementia (n=76): 41.9 (17.2)
No Stroke or Dementia (n=211):
49.9 (20.5), p<0.01 both for AD, and for All-cause dementia, versus No stroke or dementia, t-test
Beta coefficient (SE), p-value, OR (95% CI) diagnosis of dementia:
All-cause dementia diagnosis:
Univariate model (n=76): 0.96 (0.30), p=0.001, 2.6 (1.45, 4.65)
Final model (n=70): 0.79 (0.34),
p=0.02, 2.21 (1.13, 4.32)
Alzheimer’s disease diagnosis:
Univariate model (n=41): 1.05 (0.40), p=0.01, 2.86 (1.31, 6.25)
Final model (n=37): 0.98 (0.51),
p=0.05, 2.65 (0.99, 7.16)
Logistic regression, all models adjusted for home care center; Final models adjusted for age, sex, race, BMI, education, kidney function, multivitamin use, season, diabetes, hypertension, plasma homocysteine, ApoE allele status.
El-Ghoneimy 2009(37) / Inclusion: diagnosed with multiple sclerosis according to International Panel on Diagnosis of Multiple Sclerosis "McDonald" criteria and its revision.
Exclusion: age <10 or >40, metabolic or endocrinal troubles and other medical conditions, smoking. / PASAT
Faces symbol (90, T) / For all cognitive tests:
p-value not significant, Pearson correlation
Hii, 2004(31) / Inclusion: all patients admitted to geriatric rehabilitation center in Melbourne; with low trauma fractures during a 6 month period from May to November 2000; age >65.
Exclusion: patients with fractures associated with malignancy, patients taking vitamin D supplements and calcitrol on admission. / MMSE / p=0.118, simple regression.
Houston, 2007(38) / Inclusion: age 65 to 102, from two towns in Chianti geographic area of Italy, community dwelling.
Exclusion: participants who did not undergo blood tests (n=100), were missing values for serum 25OHD, or lacked both of the physical performance measures of interest. / MMSE / Men: <25 (n=59): 23.6 (0.5 SE); 25-50 (n=163): 25.6 (0.3 SE);
≥50 (n=213): 25.2 (0.3 SE), p=0.37.
Women: <25 (n=156): 23.9 (0.3 SE); 25-50 (n=249): 24.0 (0.2 SE);
≥50 (n=136): 25.0 (0.3 SE), p=0.01, adjusted ANOVA.
Lee, 2009(39) / Inclusion: non-institutionalized men, age 40-79 from 8 European countries.
Exclusion: participants missing cognitive and/or 25OHD data (missing blood samples or assay failures). / ROCF (copy)
ROCF (recall)
CTRM
DSST / Beta coefficient (95%CI): (a) 25OHD level (per 10 nmol/L); (b)25OHD categories, compared to≥75.
(a)Crude: 0.096 (0.049, 0.144), p <0.01; Adjusted: 0.064 (-0.011, 0.138), p >0.05
(b) 50-75: -0.411 (-0.869, 0.047); 25-50: -0.329 (-0.729, 0.071);
<25: -6.614 (-1.320, 0.092), all p >0.05.
(a)Crude: 0.041 (-0.027, 0.109), p >0.05; Adjusted: -0.021 (-0.163, 0.121), p >0.05
(b) 50-75: -0.116 (-0.574, 0.343); 25-50: 0.260 (-0.660, 1.180);
<25: -0.462 (-1.374, 0.451), all p >0.05.
(a)Crude: 0.075 (0.026, 0.124), p <0.01; Adjusted: -0.001 (-0.146, 0.144), p >0.05
(b) 50-75: -0.143 (-0.752, 0.465); 25-50: 0.084 (-0.874, 1.043);
<25: -0.125 (-1.304, 1.054), all p >0.05.
(a)Crude: 0.318 (0.235, 0.401), p <0.01; Adjusted: 0.152 (0.051, 0.253), p <0.01
(b) 50-75: -0.759 (-1.313, -0.204), p <0.05; 25-50: -0.768 (-1.822, 0.287), p >0.05;
<25: -1.404 (-2.681, -0.127), p <0.05.
Age by 25OHD: p interaction = 0.01, linear regression.
Llewellyn, 2009(28) / Inclusion: adults living in private households; adults age ≥65 residing in institution in England. / AMT
Cognitive impairment:
(AMT <8) / Unweighted; weighted
<25 (n=274): 8.31 (1.62); 7.97 (2.06)*; 25-50 (n=769): 8.95(1.36); 8.69 (1.97)*
≥50 (n=723): 9.34 (1.07); 9.16 (1.70)*.
Q1: Impaired (n=100), Normal (n=358); Q2: Impaired (n=55), Normal (n= 398);
Q3: Impaired (n=32), Normal (n=391); Q4: Impaired (n=25), Normal (n=407),
p<0.001 for group differences between normal and impaired, t-test.
OR (95% CI) cognitive impairment with either Q4 or 25OHD ≥75 as the reference:
Unadjusted model: Q1: 4.17 (2.56-6.78); Q2: 1.98 (1.12-3.34);
Q3: 1.29 (0.74-2.25), p linear trend <0.001.
Fully adjusted models:
Q1: 2.28 (1.36-3.83); Q2: 1.43 (0.84-2.42); Q3: 1.09 (0.64-1.86), p linear trend=0.001.
<25: 2.72 (1.48, 5.00); 25-50: 1.37 (0.81, 2.31); ≥50-<75: 0.89 (0.50, 1.56), p linear trend <0.001, logistic regression.
Liu, 1997(27) / Inclusion: age >65, medically stable, resided in long-term care facility for at least 6 months.
Exclusion: conditions known to interfere with vitamin D metabolism, such as hepatic failure, renal insufficiency (creatinine >180 mcmol/L); use of phenytoin, barbiturates, vitamin supplements, and systemic glucocorticoids; acute medical illness or not expected to survive 6 months. / SPMSQ / Measure 1 (Sept):
<25 (n=14): 6.07 (3.32)*; 25-50 (n=95): 4.82 (3.54)*; ≥50 (n=46): 5.02 (3.71)*
Measure 2 (March):
<25 (n=28): 4.46 (3.61)*; 25-50 (n=93): 5.23 (3.53)*; ≥50 (n=34): 4.79 (3.69)*
OR (95% CI) hypovitaminosis D associated with SPMSQ score ≤6:
0.90 (0.47, 1.72), univariate analysis (not significant when subsequently entered into forward conditional logistic regression model)
McGrath, 2007(40) / Inclusion: age ≥12, from one of 3 ethnic groups (non-Hispanic white, non-Hispanic black, Mexican-American), community-dwelling.
Exclusion: 'Other' ethnicities (not classified as non-Hispanic white, non-Hispanic black, Mexican-American; n=788); those aged 17-19 years who had no psychomotor tests (n=995); outlying 25OHD value (n=1 with value of 4001 nmol/L outside the range of 8.7 to 243.6 nmol/L). / Mean reaction time
DSST (coding speed)
Serial digit learning
(trials to criterion)
Memory and learning
Symbol digit substitution (errors ≥2)
Serial digit learning
(total errors ≥1) / Adjusted mean (SE) for test scores:
Adult group:
(n=4,747): Q1:229.00 (2.89); Q2: 231.19 (2.26); Q3: 235.41 (3.66);
Q4: 232.98 (1.68); Q5: 233.74 (2.57), p=0.60.
(n=4,688): Q1: 2.62 (0.04); Q2: 2.65 (0.04); Q3: 2.63 (0.03);
Q4: 2.64 (0.03); Q5: 2.69 (0.03), p=0.40.
(n=4,584): Q1:4.5 (0.1); Q2: 4.6 (0.1); Q3: 4.6 (0.1);
Q4: 4.5 (0.1); Q5: 4.7 (0.1), p=0.60.
Elderly group:
(n=4,809): Q1: 6.5 (0.1); Q2: 6.4 (0.1); Q3: 6.6 (0.1);
Q4: 6.6 (0.1); Q5: 6.4 (0.1), p=0.02, linear regression.
Adjusted OR (95% CI) of poor test score:
Adult group:
(n=4,702): Q1: 0.75 (0.52, 1.09); Q2: 0.67 (0.47, 0.96); Q3: 1.00 (0.72, 1.40);
Q4: 0.80 (0.58, 1.11), p=0.11
(n=4,584): Q1: 1.03 (0.61, 1.74); Q2: 1.21 (0.69, 2.14); Q3: 1.21 (0.77, 1.90);
Q4: 0.99 (0.62, 1.58), p=0.80, logistic regression.
Ogihara, 1990(41) / Inclusion: elderly Japanese female subjects; all were hospitalized and received about 600 mg of Ca per day in hospital.
Exclusion: no serious disease such as cardiovascular diseases, renal failure, or diabetes mellitus; none received any drug that might alter Ca metabolism. / Dementia Screening Scale of Hasegawa score
Further classified according to Ischemic score
(SDAT, VTD) / SDAT (n=22): 17.7 (8.5)
VTD (n=20): 20.7 (8.2)
No dementia (n=18): 20.7 (8.2)
p>0.05 among groups, t-test
Oudshoorn, 2008(24) / Inclusion: consecutive patients referred to the geriatric outpatient clinic of the Erasmus University Medical Center.
Exclusion: taking any vitamin suppletion at the time of investigation; serum vitamin B1, B6, B12 and 25OHD 3 levels not available. / MMSE / <50 (n=141): 18.5 ± 6.0
≥50 (n=84): 21.5 ± 6.1, p=0.003.
<25 (n=34): 16.99 ± 5.50*
25-50 (n=107): 18.86 ± 5.91*
Beta coefficient: Crude: 0.08, p<0.001; Adjusted: 0.05, p=0.01, linear regression.
Perez-Llamas, 2008(26) / Inclusion: living in one of three public nursing homes from urban area, had a normal diet.
Exclusion: enteral or parenteral nutrition, or ate pureed foods; severely cognitively impaired; severely physically impaired. / SPMSQ / <50 (n=50): 7.9 (2.0)
≥50 (n=36): 7.7 (2.0), p=0.702, t-test.
<25 (n=22): 8.55 (1.63)*
25-50 (n=28): 7.3 (2.2) *
Pearson’s correlation coefficient: no correlation, p not reported.
Przybelski, 2007(29) / Inclusion: patients presenting to clinic, referred for assessment of memory loss or behavioural problems related to cognitive impairment; measurement of serum 25OHD and vitamin B12 concentration obtained on same day as MMSE (determined on first visit for memory problems) testing.
Exclusion: Serum 25OHD, B12, and MMSE not determined on first visit for memory problems. / MMSE / <25 (n=3): 17 (5.2)*
25-50 (n=13): 16.2 (6.0)*
≥50 (n=16): 22.1 (3.7) *
R2: 0.225, p=0.006, linear regression.
Ravaglia, 1998(45) / Inclusion: non-hospitalized women, age >90 years; living in long-stay elderly facilities or at home, in Bologna, northern Italy.
Exclusion: clinical evidence of acute or chronic disease; prescribed medication for treatment of chronic disorders (including anti-inflammatory drugs, steroids and vitamin supplements). / DSM-III-R / Mean (SD) 1,25OHD concentration in pmol/L:
Dementia (n=15): 43.2 (20.16)
No dementia (n=12): 72.72 (18.96)
p=0.009, t-test
p=0.004, ANOVA adjusted for age, education.
Sakuma, 2006(42) / Inclusion: visited general hospital on Sado Island from Jan-Dec 2004 due to acute hip fracture; living on island; admitted to hospital; fracture osteoporotic in nature.
Exclusion: patients with traumatic fracture, generally poor medical condition, immediately moved to another ward for treatment of an internal disease. / Dementia severity criteria (DI-DIV) of the long-term care insurance system developed by the Japanese Ministry of Health, Labour and Welfare, 1993 / DI: 47.4 (12.5)
DII: 38.7 (12.5)
DIII-IV: 34.9 (15.0)
Cognitively normal: 53.7(17.5)
p<0.05, Kruskal-Wallis test
Wilkins, 2009(23) / Inclusion: age >55, ability to ambulate independently, ability to complete all assessments, willingness to provide serum.
Exclusion: moderate or severe cognitive impairment; history of stroke; renal failure; Parkinson's disease; use osteoporosis medications; use prescription vitamin D supplements or non-prescription vitamin D supplements greater than 800 IU daily. / MMSE
Short blessed test / <50 (n=31): 25.39 (3.9); ≥50 (n=29): 25.07 (4.9), p=0.783, t-test
African Americans only:
<50 (n=23): 24.65 (4.1); ≥50 (n=7): 29.14 (0.9), p=0.008, t-test
<50: 10.87 (7.5); ≥50: 6.31 (6.7), p=0.016, t-test
African Americans only:
<50: 11.78 (7.9); ≥50: 2.29 (2.4), p=0.004, t-test
Adjusted Beta coefficient: <50: 2.51, p=0.036, linear regression.
Wilkins, 2006(43) / Inclusion: age >60; ambulatory; able to complete all assessments; willingness to provide serum.
Exclusion: moderate or severe dementia; history of stroke; renal failure; Parkinson disease; use of prescription vitamin D supplements or nonprescription vitamin D greater than 800 IU daily. / MMSE
CDR sum of boxes
Short blessed test
Factor score
CDR >0 (dementia) / ≥50 (n=34): 26.12 (3.46); 25-50 (n=33): 26.06 (4.11); <25 (n=13): 24.77 (4.32), p=0.53.
≥50: 1.65 (2.08); 25-50: 2.36 (56); <25: 3.08 (3.07), p=0.047.
≥50: 4.15 (4.75); 25-50: 7.42 (4.80); <25: 9.92 (8.94), p=0.013.
≥50: -1.01 (1.56); 25-50: -1.35 (1.60); <25: -1.59 (1.96), p=0.52, one-way ANOVA.
Adjusted OR (95% CI) dementia:
<25 versus ≥50: 2.8 (0.64, 12.28); 25-50 versus ≥50: 1.78 (0.61, 5.19), p=0.3391 for vitamin D category, logistic regression.
Woo, 1991(44) / Inclusion: all female residents of an old age hostel; criteria of entry into hostel includes age ≥60, inability to afford own private housing, independent mobility, and activities of daily living.
Exclusion: breast cancer, leukaemia, hyperparathyroidism. / Mental test score
(adapted from Hodkinson) / Mean (SE) 25OHD concentration:
Score <4 (n=20): 49.9 (7.0)
Score ≥4 (n=60): 59.7 (2.5), p>0.05, t-test.
Case-control studies
Evatt, 2008(46) / Inclusion: AD patients free of other neurological diagnoses; controls with no neurologic disease and MMSE score ≥25.
Exclusion: history of stroke, transient ischemic attack, restless legs syndrome, essential tremor; related to any previous participant in the CRIN database groups. / NINCDS-ADRDA (AD) / AD (n=97): 86.86 (38.44)
Controls (n=99): 92.352 (36.192)
p=0.30, t-test.
Ferrier, 1990(47) / Inclusion: controls with no history of major psychiatric and/or neurological illness, MTS score ≥36, and living at home.
Exclusion: acute intercurrent illness; epilepsy; recent fracture; on the following drugs: neuroleptics, antidepressants, anticonvulsants, diuretics, hormonal treatments (e.g. steroids,oestrogens, etc.), calcium or vitamin D supplements. / DSM-III
MTS score <30
HII score <7
(ATD) / ATD (n=15): 32 (9)
Controls (n=11): 40 (25)
p>0.05
Jorde, 2006(48) / Inclusion: age >29, living in the municipality of Tromso, participated in the second phase of the 4th Tromso study or became 30, 40, 45, 60 or 75 years old during 2001; in the 5th Tromso study and at follow-up examination fulfilled criteria for SHPT without renal failure (serum calcium <2.40 mmol/L and PTH >6.4 pmol/L, normal serum creatinine and kidney function) or normal serum calcium, creatinine, and PTH values.
Exclusion: age >80; history of coronary infarction, angina pectoris, stroke; participation in other follow-up studies; having hospital records indicating serious diseases not reported in the questionnaire. / Digit Span, forward
Seashore rhythm
Trails A (seconds)
Stroop (1, 2)
DSST
CalCAP
Verbal recall
Visual recall
Word list
COWA
Stroop (3)
Trails B (seconds)
WAIS vocabulary / Beta coefficient; t value:
0.17; 1.39
-0.13; -0.90
0.18; 1.57
0.12; 1.05
-0.06; -0.74
0.04; 0.32
0.08; 0.69
0.13; 1.21
-0.03; -0.26
0.13; 1.08
-0.07; -0.68
-0.06; -0.51
0.04; 0.31, all t values not significant, multiple linear regression.
Kipen, 1995(49) / Inclusion: female, age >60, independently mobile, community-dwelling.
Exclusion: history of rheumatoid arthritis, hyperparathyroidism, fracture; taking estrogens, corticosteroids, or vitamin D; alcohol intake greater than 40 g/day; renal, liver, or metabolic bone disease or abnormalities. / DSM-III-R
NINCDS-ADRDA
(Mild dementia) / Mild dementia (n=20): 61 (33)
Controls (n=40): 90 (38)
p=0.003, t-test.
Luckhaus, 2009(50) / Exclusion: positive history of osteoporosis or non-traumatic fractures or positive to risk factors of osteoporosis (identified through 45-item checklist inquiring about relevant risk factors). / NINCDS-ADRDA (AD)
Petersen, 1999 (MCI) / AD (n=20): 39.5 (24.5); MCI (n=19): 48 (18); Controls (n=8): 36 (13)
p>0.05 for all comparisons, two-sided ANOVA (across groups with sex as a covariate), Mann Whitney U-test (inter-group comparisons).
Martyn, 1989(51) / Inclusion: all female patients with AD in one Hospital under care of one psychogeriatrician; controls were all in-patients resident in same wards as cases, with chronic psychiatric conditions (schizophrenia, uni- or bipolar depressive illness, or long-term mental handicap); in-patients for >6 weeks.
Exclusion: hypothyroidism, vitamin B12 deficiency, taking anticonvulsant or other drugs known to affect calcium metabolism, bone fracture within previous 12 months. / Intellectual deterioration over a period of more than 6 months and no indication of any other cause for dementia / AD (n=27): 28.70
Controls (n=34): 35.69
difference (95% CI): -2.8 (-4.9, -0.7) p<0.05
Nes, 1988(52) / Inclusion: random sample of persons age >75 from socio-medical survey of independent-living elderly.
Exclusion: living in institution (hospital, nursing home). / DSM-III / Dementia (n=15): 50.42 (24.21); Controls (n=15): 58.16 (20.97)
p>0.05, χ2 test or Mann Whitney
Not taking vitamin D supplements:
Dementia (n=5): 35.4; Controls (n=11): 53.16
p>0.05, χ2 test or Mann Whitney
Sato, 2005(53) / Inclusion: ambulatory female patients, age >70; cognitively normal controls without fractures (as determined by radiography) at any sites.
Exclusion: impairment of renal, cardiac, or thyroid function or have known causes of osteoporosis such as hyperparathyroidism; treatment with corticosteroids, estrogens, calcitonin, bisphosphonate, calcium, menatetrenone, or vitamin D for 3 months or >12 months preceding study, or administered these agents for even brief period in preceding 2 months. / DSM-III-R
MMSE / Severe AD (n=58): 25 (10)
Mild AD (n=42): 40.5 (10)
Controls (n=100): 60.5 (11)
p<0.001 among groups, one-way ANOVA.
Sato, 1998(54) / Inclusion: patients admitted to nursing home in Japan in July or August of 1994, ambulatory, age >70; cognitively normal controls without fractures.
Exclusion: patients diagnosed with conditions such as osteoporosis, received any drugs known to alter bone metabolism; controls with evidence of vertebral fractures or showed known causes of osteoporosis such as hyperparathyroidism or renal osteodystrophy; impairment of renal, cardiac, or thyroid function; history of therapy with corticosteroids, estrogen, calcitonic, etidronate, calcium, or vitamin D for 3 months or longer during 18 months preceding study, or treatment during 2 months immediately preceding study. / DSM-III-R
NINDS-ADRDA / AD (n=46): 17.7 (9.7)
Controls (n=140): 53.91 (7.7)
p<0.0001, t-test.
Walker, 2009(55) / Inclusion: age 45, menopausal for at least 1 year, English speaking; PHPT patients had hypercalcemia and elevated or inappropriately normal PTH level.
Exclusion: male sex; malignancy other than nonmelanoma skin cancer; alcoholism; significant liver or kidney disease; illnesses requiring more than 2 weeks of corticosteroids per year; traumatic brain injury or neurosurgery, seizure disorders, untreated mood disorders; use of medications known to affect mineral metabolism. / NAART
WMS logical memory (Russell revision)
SRT
RVDLT
BCT (Victoria revision)
RTD
WAIS-R DSST
WAIS digit span / For all cognitive tests:
No linear association observed between any cognitive variable that was abnormal at baseline and serum 25OHD concentration;
No relationship observed between changes in cognitive variables and change in serum 25OHD concentration.
Cohort studies
Llewelyn, 2010(57) / Inclusion: age 65 and older, from Greve, Chianti or Bagno a Ripoli, Tuscany Italy, completed at least one follow-up cognitive assessment between 1988 and 2006. / MMSE
Trails A (seconds)
Trails B (seconds)
Dementia
Cognitive decline:
MMSE (decline of ≥3pts)
Trails A (seconds)
Trails B (seconds)
(worst 10% of decline or
test discontinued)
Cognitive decline:
MMSE / Baseline:
<25 (n=175): 23.7 (5.3); 25-50 (n=360): 25.2 (3.3); ≥50-75 (n=166): 26.0 (3.0);
≥75 (n=157): 26.3 (2.8); p<.001; 2-way interaction: p=.76.
<25:151.8 (94.7); 25-50: 114.9 (75.5); ≥50-75: 94.2 (69.7); ≥75: 87.2 (62.1); p<.001; 2-way interaction: p=.40.
<25: 239.9 (78.4); 25-50: 219.6 (80.8); ≥50-75): 188.5 (32.4); ≥75: 180.5 (87.0); p<.001; 2-way interaction: p=.91, one-way ANOVA.
Number (%):
<25: 16 (9.1); 25-50: 10 (2.8); ≥50-75: 2 (1.2); ≥75: 1 (0.6); p<.001, χ2 test.
Adjusted RR (95% CI) 6-year substantial cognitive decline, with >75 as reference:
≥50-75: 1.19 (0.84, 1.58); 25-50: 1.09 (0.78, 1.43); <25: 1.60 (1.19, 2.00); p trend=.02
≥50-75: 0.95 (0.55, 1.51); 25-50: 1.25 (0.75, 1.71); <25: 1.16 (0.65, 1.84); p trend=.44
≥50-75: 0.99 (0.74, 1.23); 25-50: 1.11 (0.88, 1.32); <25: 1.31 (1.03, 1.51); p trend=.04, logistic regression. Same pattern of association observed when sample restricted to baseline nondemented participants or after further adjusted for potential mediators stroke, hypertension, diabetes.
Adjusted beta value (SE) year-on-year change in MMSE-measured cognitive function, with >75 as reference:
All participants:
S (≥50 to <75): -0.111 (0.115); I: -0.035 (0.095); D: -0.321 (0.109), p trend =.03.
Nondemented participants only:
S (≥50 to <75): -0.130 (0.114); I: -0.051 (0.096); D: -0.310 (0.109), p trend =.04.
Random-effects models.
Slinin, 2009(32) / Inclusion: age >65; able to consent; able to walk without the assistance of another person; not had bilateral hip replacement; able to provide self-reported information; expected to reside near the clinical site for the duration of the study; had no condition that in the judgment of the site investigator would make the individual unable to participate or survive the duration of the study, or for whom participation would be inappropriate; be among the random sample of 1,606 providing a blood sample at MrOS study baseline. / Prevalent CI:
3MS <80
Trails B (time <1.5 SD
above the mean)
Incident CI:
3MS (baseline <80 or
decline of ≥5 points on
follow-up)
Trails B (change in
completion time 1 SD or
more above sample's
mean change in
completion time for those
without prevalent
impairment at baseline (>
50.7 sec), baseline to
follow-up) / Adjusted OR (95% CI) cognitive impairment at baseline:
Model 1 (n=1,604): Q1: 1.84 (0.81, 4.19); Q2: 1.41 (0.61, 3.28);
Q3: 1.18 (0.50, 2.81); Q4: referent, p trend =0.12.
Multivariable Model (n=1,598): Q1: 0.93 (0.36,2.39); Q2: 1.24 (0.49, 3.13);
Q3: 0.98 (0.38, 2.52); Q4: referent, p trend =0.97.
Model 1 (n=1,564): Q1: 1.66 (0.98, 2.82); Q2: 0.96 (0.54, 1.69);
Q3: 1.30 (0.76, 2.22); Q4: referent, p trend =0.12.
Multivariable Model (n=1,559): Q1: 1.09 (0.61, 1.93); Q2: 0.81 (0.45, 1.48);
Q3: 1.14 (0.65, 2.00); Q4: referent, p trend =0.96, logistic regression
Adjusted OR (95% CI) cognitive impairment at follow-up with Q4 referent:
Model 1 (n=1,138): Q1: 1.53 (0.99, 2.37) Q2: 1.30 (0.86, 1.97);
Q3: 1.08 (0.71, 1.64), p trend =0.04.
Multivariable Model (n=1,136): Q1: 1.41 (0.89, 2.23); Q2: 1.28 (0.84, 1.95);
Q3: 1.06 (0.70, 1.62), p trend =0.10.
Model 1 (n=1,053): Q1: 1.02 (0.52, 2.03); Q2: 1.05 (0.56, 1.97);
Q3: 1.01 (0.54, 1.89), p trend =0.91.
Multivariable Model (n=1,051): Q1: 1.08 (0.53, 2.19); Q2: 1.08 (0.57, 2.04);
Q3: 0.96 (0.50, 1.82); p trend =0.75, logistic regression. AF: age, clinic site, season of blood draw (model 1); age, clinic site, season of blood draw, race/ ethnicity, education, self-reported health status, IADL impairments, smoking, alcohol consumption, BMI, physical activity (multivariate model).

Randomized controlled trial studies

Chandra, 2001(59) / Inclusion: apparently healthy volunteers, independently living, age >65, from middle socioeconomic class with average family income of C$48,000 per annum; no chronic or serious acute illness, specifically any form of psychiatric illness or dementia; have not taken any nutritional supplements in 3 months before recruitment; not taking other medications that might have interfered with nutrition status, cognitive function. / MMSE
Wechsler memory Halstead–reitan categories
Buschke consistent long-term retrieval
Digit span (forward)
Salthouse listening span
Long-term memory recall / For all cognitive tests:
All participants combined: no significant correlation between levels of individual nutrients and cognitive function test scores, p-value not reported, Pearson’s coefficient;
No single nutrient appeared to influence cognition, p-value not reported, multiple regression
Corless, 1987(58) / Inclusion: long-stay patients, plasma 25OHD <16 ng/ml, plasma potassium not less than 3.3 mmol/l; plasma creatinine not greater than 150 micromol/l, thought likely to remain an inpatient for a further 8 weeks.
Exclusion: overt osteomalacia; recently prescribed calciferol. / Mental assessment score
(derived from Silver test of mental ability orientation questions, and practical tests) / No significant difference between treatment (n=41) and control (n=41) groups, p-value not reported, test not reported.
Manders, 2009(33) / Inclusion: age ≥60, institutionalized for at least 2 months, MMSE score ≥10 points and BMI ≤30 kg/m2, from 9 different nursing homes in the Netherlands.
Exclusion: serious morbidity (malignant cancer, severe infectious diseases, use of parenteral food or structural use of tube feeding, absorption disorders, terminal cancer) or interfering co-interventions. / MMSE*
Verbal fluency
(semantic categories)
ADAS-cog / <25 (n=106): 19.4 (5.5)*; 25-50 (n=54): 21.9 (5.3)*; ≥50 (n=8): 21.5 (3.8)*
25-50 versus <25: p=0.018, Bonferroni test*
Differences between changes during 24 weeks in supplement and placebo groups:
Animals category:
Supplement (n=78): 0.6 (3.4)
Placebo (n=33): 0.0 (3.2), p=0.410
Professions category:
Supplement (n=78): -0.1 (3.0)
Placebo (n=33): -0.3 (2.6), p=0.687
Median (p10, p90):
Total score:
Supplement (n=78): 0.0 (-5.0, 8.0)
Placebo (n=33): 1.0 (-5.6, 6.0), p =0.845
Memory/orientation component:
Supplement (n=78): 0.0 (-4.8, 5.0)
Placebo (n=33): 0.0 (-3.0, 5.0), p =0.792
Language component:
Supplement (n=78): 0.0 (-1.0, 2.7)
Placebo (n=33): 0.0 (-2.6, 3.0), p=0.884
Praxis component:
Supplement (n=78): 0.0 (-1.0, 2.0)
Placebo (n=33): 0.0 (-2.0, 3.6), p=0.703
Vitamin D concentration:
Supplement (n=78): 21.0 (-2.7, 57.9)
Placebo (n=33): -0.5 (-5.4, 24.8), p<0.001, Mann-Whitney U-test
Before-after with comparison group
Przybelski, 2008(56) / Inclusion: individuals residing in one of eight nursing homes in northeast or south-central Wisconsin for ≥6 months; capable of ambulation (independently or with assistance) and able to answer questions; no calcium or vitamin D limitation specified in study protocol.
Exclusion: a history of renal failure (serum creatinine >2.0 mg/dl) diagnosis of liver failure; known malabsorption (e.g., celiac sprue or radiation enteritis); known disorders of parathyroid function, hypercalcemia, hypocalcemia or other abnormalities of calcium or phosphate metabolism; known history of vitamin D intoxication and granulomatous diseases (e.g., sarcoidosisor tuberculosis). No calcium or vitamin D intake limitation was specified in the study protocol. / CDT
Verbal fluency
(semantic) / Treatment group (n=25) 4 weeks: 5.7 (0.5 SE)
Comparison group (n=38) 4 weeks: 5.7 (0.5 SE)
Treatment group 4 weeks: 8.3 (1.2 SE)
Comparison group 4 weeks: 8.3 (0.8 SE)
p-value not reported, no significant differences, repeated-measures ANOVA.

1 Unless specified otherwise.