Vitamin B6, B12, and Folic Acid Supplementation and Cognitive

REVIEW ARTICLE

Vitamin B6,B12, and Folic Acid Supplementation and Cognitive Function A Systematic Review of Randomized Trials

Ethan M. Balk, MD, MPH; Gowri Raman, MD; Athina Tatsioni, MD; Mei Chung, MPH; Joseph Lau, MD; Irwin H. Rosenberg, MD

Background: Despite their important role in cognitive in cognitive function in people with cognitive impair- function, the value of B vitamin supplementation is un- ment and low baseline serum folate levels. Six trials of known. A systematic review of the effect of pyridoxine hy- combinations of the B vitamins all concluded that the in- drochloride (hereinafter “vitamin B6”), cyanocobalamin or terventions had no effect on cognitive function. Among hydroxycobalamin (hereinafter “vitamin B12”), and folic acid 3 trials, those in the placebo arm had greater improve- supplementation on cognitive function was performed. ments in a small number of cognitive tests than participants receiving either folic acid or combination B- Methods: Literature search conducted in MEDLINE with vitamin supplements. The evidence was limited by a supplemental articles from reviews and domain ex- sparsity of studies, small sample size, heterogeneity in perts. We included English language randomized con- outcomes, and a lack of studies that evaluated symp- trolled trials of vitamins B6 and/or B12 and/or folic acid toms or clinical outcomes. supplementation with cognitive function outcomes. Conclusion: The evidence does not yet provide ad- Results: Fourteen trials met our criteria; most were of equate evidence of an effect of vitamin B6 or B12 or folic low quality and limited applicability. Approximately 50 acid supplementation, alone or in combination, on cog- different cognitive function tests were assessed. Three tri- nitive function testing in people with either normal or als of vitamin B6 and6ofvitaminB12 found no effect over- impaired cognitive function. all in a variety of doses, routes of administration, and populations. One of 3 trials of folic acid found a benefit Arch Intern Med. 2007;167:21-30

LOBALLY,24MILLION been proposed to prevent or reverse cog- people have some form nitive decline. of dementia, with 4.6 This systematic review examines million new cases diag- whether supplementation with pyridox- nosed each year.1 It is ine hydrochloride (hereinafter “vitamin estimatedG that the number of people B6”), cyanocobalamin or hydroxycobal- affected will double every 20 years and amin (hereinafter “vitamin B12”), and folic reach 81 million by 2040. Pharmaco- acid can prevent, decrease the progres- therapy of Alzheimer disease and other sion rate of, or reverse the neurologic changes associated with age-related neu- CME course available at rodegenerative conditions such as Alz- Author Affiliations: Tufts-New www.archinternmed.com heimer disease in humans. The review England Medical Center was conducted as part of a larger evidence Evidence-based Practice Center, dementias can provide only modest cog- report on the association between B vita- Institute for Clinical Research nitive or disease-modifying benefits.2 mins and berries and neurocognitive and Health Policy Studies, However, even modest benefits may have function.6 Tufts-New England Medical significant effects on quality of life, car- Center (Drs Balk, Raman, egiver burden, and societal economic METHODS Tatsioni, and Lau and costs. Increased homocysteine levels in Ms Chung), and Nutrition and conjunction with low levels of folate, Neurocognition Laboratory, STUDY ELIGIBILITY Jean Mayer US Department of vitamin B6, and vitamin B12, which inter- Agriculture Human Nutrition act to control homocysteine, have been We conducted a comprehensive literature Research Center on Aging at reported to correlate with decreased per- search for publications on B vitamins in Tufts University formance on cognitive tests.3-5 For these MEDLINE and Commonwealth Agricultural (Dr Rosenberg), Boston, Mass. reasons, B vitamin supplementation has Bureau (CAB) Abstracts from inception through

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 21

Downloaded From: https://jamanetwork.com/ on 10/01/2021 ity studies were susceptible to some bias. They had some deficiencies but none 6914 MEDLINE, CAB Abstracts Electronic Searches 16 Reference Lists of Studies, Reviews, and likely to cause major bias. (3) Poor qual- Domain Expert Suggestions ity studies had significant bias. They had serious errors in design, analysis, or re- 6631 Failed to Meet Criteria for Any Topic in Abstract porting or may have had a large amount of missing information or discrepan- cies in reporting. Each included study 299 Total Articles Retrieved 283 Electronic Searches was graded by at least 2 people. When 16 References and Domain Experts there were disagreements, additional re- viewers graded the studies, and consen-

20 Review Articles sus was reached. 78 Animal or In Vitro Studies APPLICABILITY GRADE 201 Human Studies Retrieved 185 Electronic Searches 16 References and Domain Experts Applicability addresses the relevance of a given study that is distinct from the question of the study’s methodologic 146 Studies of Correlation Between B Vitamin Level quality. The applicability of a study is or Dietary Intake and Cognitive Function 32 Not Randomized Controlled Trial dictated by the questions and popula- 9 Combination Vitamin B and Other Interventions, tions that are of interest to those ana- Vitamin B , or No Outcomes of Interest 1 lyzing the studies. We categorized studies within a tar- 14 Studies Reviewed, 18 Comparisons get population into 1 of 3 levels of appli- 6,7 3 Vitamin B6 cability : (1) In widely applicable stud- 6 Vitamin B12 3 Folate ies, the sample was representative of 6 Combination B Vitamins individuals at increased risk for, or diag- nosed as having, age-related neurocognitive disorders. They included both sexes, Figure. Flow diagram of study selection process. CAB indicates Commonwealth Agricultural Bureau. an appropriate age range, and other important features of the target population February 2, 2005; an update limited to dementia lacking separate analyses for (eg, general health status). At least 30 par- randomized trials and systematic re- disease types. We also excluded stud- ticipants were analyzed. (2) For moder- views was performed on August 17, ies of “multivitamins” that included vi- ately applicable studies, the sample was 2006. Search terms included the com- tamins other than B vitamins. representative of a relevant subgroup of mon and chemical names for the B vi- the target population but not the entire tamins (folic acid, folate, pteroyl- STUDY SELECTION population. Limitations included such fac- glutamic, folacin, riboflavin, lactoflavin, AND DATA EXTRACTION tors as a narrow age range and a setting thiamin(e), cobalamin, cyanocobalamin, that applied to only a portion of the gen- pyridoxine, pyridoxal, pyridoxamine, and All citations identified through the lit- eral population (eg, nursing home). At vitamins B , B , B , and B ) and neu- 1 2 6 12 erature search were screened accord- least 10 participants were analyzed. (3) In rocognitive terms including nervous sys- ing to the eligibility criteria, and re- narrowly applicable studies, the sample tem diseases, cognitive disorders, de- trieved articles were evaluated against the was representative of a narrow subgroup lirium, amnestic, neurodegeneration, same criteria. Each accepted study was of participants only and was of limited ap- dementia, Alzheimer, Lewy body, brain, extracted by a single reviewer, and a sec- plicability to other subgroups. neuron, and nerve cell. Additional stud- ond reviewer independently verified the On review of the literature, we de- ies were sought from neurology and vi- extracted data. Data extraction issues termined that studies of short-term B vitamin research experts and from refer- were resolved by consensus. tamin supplementation (ie, 1 or 3 ence lists of selected articles, review months), but longer-term cognitive test- articles, and meta-analyses. METHODOLOGIC ing (ie, at 3 or 12 months), were of lesser Study eligibility criteria included relevance to the question of whether B peer-reviewed reports of randomized QUALITY GRADE vitamin supplementation should be rec- controlled trials of clearly defined B vi- ommended to prevent cognitive de- tamin supplements (B6,B12, and folic We used a 3-category grading system to cline. These studies were included but acid), where specific type of vitamin, denote the methodologic quality of each deemed to be of limited applicability. dose, and route of administration were trial6,7: (1) Good quality studies had the reported. We included English lan- least bias. These studies mostly ad- guage publications of trials of adult par- hered to the commonly held concepts OUTCOMES AND ticipants with outcomes related to di- of good quality: they were formal and METRICS REPORTED agnosis or severity (degree) of Alzheimer randomized with a clear description of disease, other age-related neurocogni- the population, setting, interventions, We evaluated all outcomes relevant to tive disorders, cognitive impairment, or and comparison groups; they used ap- neurocognitive function that were re- tests of cognitive function. We ex- propriate outcome measurements, sta- ported in studies. In all included trials, cluded studies of mental retardation, en- tistical and analytic methods, and re- this involved the use of one or more cog- cephalopathies, peripheral neuropathy porting conventions; and they contained nitive function tests or subtests. and other lower motor neurodegenera- no reporting errors, less than a 20% Three sets of data were evaluated: tion, subacute combined degeneration, dropout rate, clear reporting of drop- (1) the mean baseline levels in both in- vascular dementia, and mixed causes of outs, and no obvious bias. (2) Fair qual- tervention and control arms, (2) within-

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 22

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 1. Summary of Intervention Studies Evaluating the Effect of B Vitamins on Neurocognitive Outcomes

Quality, No. Applicability, No. Vitamin Studies, Participants, Supplementation No. No.* Good Fair Poor Wide Moderate Narrow Results

B6 3 145 1 1 1 0 1 2 No association

B12 6 210 1 3 2 0 2 4 No association or worse outcome† Folic acid 3 39 0 2 1 0 1 2 No association or improved outcome‡

Combination B6,B12, 6 472 2 2 2 2 2 2 No association or worse outcome§ and folic acid

Abbreviations: Vitamin B6, pyridoxine hydrochloride; vitamin B12, cyanocobalamin or hydroxycobalamin. 9 *Receiving B vitamin supplementation. For vitamins B6,B12, and folic acid, these numbers are approximate because Bryan et al reported only that 75 participants were randomized to 1 of the B vitamins or placebo. For combination treatment, the number is approximate because Clarke et al18 reported only that 128 participants with data on cognitive function were randomized between B vitamins and placebo. †Statistically significant net worsening in cognitive function reported in a small minority of cognitive tests performed. ‡One study found a benefit in people with cognitive impairment and low baseline serum folate levels. §Two studies found greater improvement in several cognitive function tests among people with normal cognitive function in the placebo group than among those in the combination B vitamin group.

Table 2. Effect of Pyridoxine Hydrochloride (Vitamin B6) Intervention on Cognitive Function Tests

Intervention (Control) Change Net Change B Vitamin Population Intervention Dose, mg/d Participants Characteristic Quality/ Maximum (Control) P P Source (Duration) (Controls), No. (Mean Age, y) Applicability Test Score* Base Value Value Value Value Value Stott et al,8 25 88/78‡ (78) Ischemic Good/narrow Telephone interview 39 25.5 (ND) −0.2 (0.0) NS (NS) −0.2 NS 2005 (3/12 vascular for cognitive mo†) disease (74) status20 Letter-digit coding ND 19.7 (ND) −0.6 (0.0) NS (NS) −0.6 NS test21 Bryan et al,9 75 (5 wk) ~19 (~19)§ Normal (74) Fair/narrow Digit-symbol coding ND 63.4 (62.3) ϩ7.3 (ϩ4.8) NS (NS) ϩ2.5 NS 2002 (WAIS)22 Verbal ND 22.1 (21.9) ϩ0.6 (−0.2) NS (NS) ϩ0.8 NS ability–vocabulary (WAIS)22 Digit span–backward ND 6.1 (7.1) 0.0 (ϩ0.4) NS (NS) −0.4 NS (WAIS)22 Stroop23 ND 2.34 (2.51) −0.06 (−0.12) NS (NS) ϩ0.06 NS Verbal fluency–initial ND 29.3 (23.7) ϩ1.2 (ϩ3.3) NS (NS) −2.1 NS letter24 Deijen et 12 (12 wk) 38 (38) Normal (83) Poor/moderate Associate recognition 93.2࿣ (3.9 ࿣) ϩ0.1 (−1.1) NS (ND) ϩ1.2 ND al,10 1992 task25 Long-term memory 9¶ 0.35 (0.45) −0.35 (ϩ0.45) ND (ND) −0.8¶ Ͻ.03 storage25

Abbreviations: ND, no data; NS, nonsignificant; Stroop, Stroop Color-Word Test; vitamin B6, pyridoxine hydrochloride; WAIS, Wechsler Adult Intelligence Scale. *Unless otherwise indicated, higher score indicates better cognitive function. †Intervention lasted 12 weeks; follow-up cognitive testing occurred at 1 year. ‡Number tested with telephone interview of cognitive status20/number tested with letter digit coding test.21

§Seventy-five participants were randomized among vitamins B6,B12, folic acid, and placebo. ࿣Results reported graphically. ¶Lower score indicates better cognitive function.

cohort changes (ie, final minus baseline retrieved. An additional 16 studies tion; 32 were other nonrandom- values), and (3) the net change of the out- were identified in review articles and ized studies; and 9 randomized trials come (change in intervention cohort mi- study reference lists or were for- did not evaluate either interven- nus control cohort). In addition, we included the reported P values of both the warded by domain experts. From tions or outcomes of interest within-cohort change and the net change. these 202 articles, 14 randomized (Figure). One trial reported sepa- We reported P value less than .10; those controlled trials of vitamins B6,B12, rate data for different age groups of above .10 and those reported as nonsig- and folic acid supplementation quali- cognitively normal women.9 After re- nificant were described as “NS.” fied for this review (these included viewing all studies, we included only 18 comparisons of B vitamins with the data from the older cohort of RESULTS placebo).8-19 Among the rejected women, aged 65 to 92 years, to be studies, 146 were studies of corre- consistent with the other trials, The literature searches yielded 6914 lations between B vitamin levels or which all included only older par- citations, of which 185 articles were dietary intake and cognitive func- ticipants. Table 1 lists the number

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 23

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 3. Effect of Cyanocobalamin or Hydroxycobalamin (Vitamin B12) Intervention on Cognitive Function Tests in Randomized Controlled Trials

Intervention B Vitamin (Control) Change Net Change Participants Population Intervention Dose (Controls), Characteristic Quality/ Maximum (Control) P P Source (Duration) No. (Mean Age, y) Applicability Test Score* Base Value Value Value Value Value Hvas et 1 mg/wk, 70 (70) Cognitive Good/narrow MMSE26 30 26 (27) ϩ0.3 (ϩ0.2) NS (NS) ϩ0.1 NS al,11 injection impairment CAMCOG27 100 89 (89) ϩ1.3 (ϩ1.9) .04 (.001) −0.6 NS 2004 (1/3 mo†) (75) 12-Word learning 12 5 (5) ϩ0.2 (ϩ0.4) NS (.04) −0.2 NS test–immediate28 12-Word learning 12 2 (2) ϩ0.2 (ϩ0.7) NS (.001) −0.5 .04‡ test–15 minutes28 Eussen 1 mg/d, oral 64 (65) Normal to Fair/moderate Digit span–forward 16 7.4 (7.6) ϩ0.1 (ϩ0.2) NS (NS) −0.1 NS et al,29 (24 wk) cognitive (WAIS)22 2006 impairment Digit span–backward 14 4.9 (4.7) −0.3 (ϩ0.6) Ͻ.05 (Ͻ.05) −0.9 .001‡ (73) (WAIS)22 Executive function 12 5.3 (4.8) ϩ0.8 (ϩ0.6) NS (NS) ϩ0.2 NS similarities (WAIS)30 Speed–finger ND§ 453 (409) −41 (−20) NS (NS) −21 NS tapping, ms31 Speed–trail-making ND§ 75.4 (72.0) ϩ2.1 (ϩ1.9) NS (NS) ϩ0.2 NS test, part A32 15-Word 75 30.9 (30.0) ϩ4.3 (ϩ5.7) NS (NS) −1.4 NS learning– immediate recall33 15-Word 15 4.8 (5.1) ϩ0.7 (ϩ1.0) NS (NS) −0.3 NS learning–delayed recall33 15-Word 30 25.9 (25.3) ϩ0.7 (ϩ1.7) Ͻ.05 (Ͻ.05) −1.0 .03‡ learning– recognition33 Executive ND§ 2.7 (2.9) ϩ0.1 (−0.1) NS (NS) ϩ0.2 NS function–trail- making test, part C/part A,32 Executive ND§ 2.2 (2.1) 0.0 (ϩ0.7) NS (NS) −0.7 NS function–Stroop test, part 3/part 234 Construction– 36 28.0 (27.7) ϩ2.0 (ϩ1.5) NS (NS) ϩ0.5 NS complex figure of Rey, copy35 Speed–motor ND§ 627 (673) ϩ20 (−55) NS (NS) ϩ75 NS planning 2, ms31 Executive ND§ 898 (1012) −35 (−22) NS (NS) −13 NS function–motor planning 3, ms31 Executive 24 15.6 (15.5) ϩ1.0 (ϩ1.0) NS (NS) 0 NS function–Raven36 Word ND 17.6 (17.4) 0.0 (−0.9) NS (NA) ϩ0.9 NS fluency–animals, No. of nouns37 Word fluency–letter, ND 16.2 (15.2) −0.7 (ϩ2.3) Ͻ.05 (Ͻ.05) −3.0 NS No. of nouns37

(continued)

of studies, number of participants, mal cognitive function or limited to after the vitamin supplementation quality, applicability, and summa- those with ischemic vascular dis- stopped.8 rized results of the evidence for the ease (Table 2).8-10 Sample sizes Across studies, among 9 differ- effects of B vitamin intervention on ranged from about 19 to 88 partici- ent cognitive function subtests used neurocognitive conditions. pants taking vitamin B6 supple- in the 3 trials, no statistically sig- ments; doses ranged from 12 to 75 nificant difference in testing was VITAMIN B6 mg/d, considerably higher than the found after vitamin B6 supplemen- US Recommended Daily Allowance tation in all but 1 test (Deijen et al10), Three trials of either good, fair, or (RDA) dose of 1.3 to 1.7 mg/d. Vita- where there was a significant im- poor quality and of narrow to mod- min B6 supplementation was given provement in long-term memory erate applicability investigated the ef- for 5 or 12 weeks; however, nota- with vitamin B6 compared with pla- fect of B6 intervention on cognitive bly, 1 study performed follow-up cebo. Among all the subtests, there function in people with either nor- cognitive testing at 1 year, 9 months was no consistent direction (net im-

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 24

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 3. Effect of Cyanocobalamin or Hydroxycobalamin (Vitamin B12) Intervention on Cognitive Function Tests in Randomized Controlled Trials (cont)

2002 (~4 wk) vitamin B12 levels¶ (82) 0.05 mg/d, 10 (11) Normal, with Fair/narrow MMSE26 30 19.7 (19.6) ϩ1.0 (ϩ1.6) ND (ND) −0.6 NS oral low serum

(~4 wk) vitamin B12 levels¶ (82) Kwok 1 mg/mo,# 27 (23) Normal, with Poor/ MMSE26 30 22.2 (23.8) ϩ0.1 (ϩ0.2) NS (NS) −0.1 NS et al,13 injection low serum moderate Digit span ND 10.4 (11.6) ϩ0.3 (−1.0) NS (NS) ϩ1.3 NS 22 1998 (3-6 mo) vitamin B12 (WAIS-R) levels**(77) Visual memory22 ND 12.7 (15.3) −3.0 (−3.7) NS (NS) ϩ0.7 NS Verbal memory22 ND 7.8 (11.4) −1.1 (−2.1) NS (NS) ϩ1.0 NS Verbal IQ22 ND 58.2 (60.1) ϩ1.1 (−1.2) NS (NS) ϩ2.3 NS Performance IQ22 ND 74.9 (84.3) ϩ5.8 (−1.5) .005 (NS) ϩ7.3 NS Kral 0.1 mg/d, 18 (22) Normal (ND) Poor/narrow Memory quotient38 100 90 (88) −3 (ϩ4) NS (ND) −7 NS et al,14 5 d/wk, Lower limit of ND 38 (45) −12 (−4) NS (NS) −8 NS 1970 injection retention39 (14 wk) Upper limit of ND 48 (45) −6 (−5) NS (NS) −1 NS retention39

Abbreviations: CAMCOG, the cognitive (COG) and self-contained part of the Cambridge Examination for Mental Disorders of the Elderly Examination (CAMDEX) (assesses orientation, language, memory, praxis, attention, abstract thinking, perception, and calculation) (includes the Mini-Mental State Examination [MMSE]); IQ, intelligence quotient; MS, millisecond; ND, no data; NS, nonsignificant; Stroop, Stroop Color-Word Test; WAIS, Wechsler Adult Intelligence Scale; WAIS-R, WAIS–Revised. *Unless otherwise indicated, higher score indicates better cognitive function. †Intervention lasted 4 weeks; follow-up cognitive testing occurred at 3 months. ‡Placebo better than B vitamin. §Lower score indicates better cognitive function.

࿣Seventy-five participants were randomized among vitamins B6,B12, folic acid, and placebo. ¶Between 136 and 203 pg/mL (100-150 pmol/L). #One milligram 3 times in week 1, then 1 mg/wk for 3 weeks, then 1 mg/mo. **Less than 163 pg/mL (Ͻ120 pmol/L).

provement vs net worsening) of ef- Two trials each recruited cogni- nitive function assessment instru- fect with supplementation. tively intact participants with nor- ments used. There is inadequate evidence to mal or low vitamin B12 levels; 1 trial For most of the cognitive tests per- support any dose effect of vitamin B6 included participants with cogni- formed, no effect was found for vi- on the outcomes. There was insuffi- tive impairment; and 1 included tamin B12. Only 3 of the 35 cogni- cient evidence from the trials regard- people with a range of cognitive func- tive function tests and subtests ing differences in effect on tests of dif- tion. The 6 trials used different doses showed a statistically significant different cognitive domains. No other and administration routes of vita- ference between participants receiv- interactions were reported in the min B12 interventions ranging from ing vitamin B12 compared with pla- studies. 0.01 mg/d orally to 1 mg/wk intra- cebo. Notably, statistically significant muscularly. Three trials used oral vi- net worsening on cognitive function- VITAMIN B12 tamin B12, and 3 used intramuscular ing tests was found in 2 studies (Hvas 11 29 vitamin B12. Doses were higher than et al and Eussen et al ). However, Six trials of variable quality assessed the US RDA of 2.4 µg/d. The dura- in 1 of these tests, the effect was the effect of vitamin B12 intervention of supplementation ranged from driven primarily by a statistically sig- tion on cognitive function in hu- 4 weeks to 6 months. Notably, Hvas nificant improvement in the pla- mans ( Table 3).9,11-14,29 All studies et al11 retested cognitive function 2 cebo arm that was not seen in the 11 had narrow to moderate applicabil- months after stopping vitamin B12 treatment arm. Neither study con- ity. Sample sizes ranged from 18 to supplementation. There was large sidered the effects to be clinically sig- 70 participants receiving vitamin B12. heterogeneity among trials in the cog- nificant. In contrast, a large nonsig-

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 25

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 4. Effect of Folic Acid Intervention on Cognitive Function Tests

Intervention B Vitamin (Control) Change Net Change Participants Population Intervention Dose, mg/d (Controls), Characteristic Quality/ Maximum (Control) P P Source (Duration) No. (Mean Age, y) Applicability Test Score* Base Value Value Value Value Value Fioravanti 15 (60 d) 16 (14) Cognitive Fair/moderate Acquisition and recall‡ ND 55.3 (62.1) ϩ4.2 (−1.2) ND (ND) ϩ5.5 Ͻ.007 et al,15 impairment, Delayed recall‡ ND 56.1 (63.0) ϩ7.4 (0.0) ND (ND) ϩ7.4 Ͻ.007 1997 low folate Memory index‡ ND 49.3 (57.1) ϩ6.8 (ϩ0.5) ND (ND) ϩ6.3 Ͻ.002 levels† (80) Encoding‡ ND 4.3 (5.3) ϩ0.5 (−0.2) ND (ND) ϩ0.7 Ͻ.005 Cognitive efficiency‡ ND 3.28 (4.25) ϩ0.63 (ϩ0.06) ND (ND) ϩ0.57 NS Attention efficiency‡ ND 6.40 (6.83) ϩ1.12 (ϩ0.07) Ͻ.05 (NS) ϩ0.97 NS Bryan 0.75 (5 wk) ~19 (~19)§ Normal (74) Fair/narrow Digit-symbol coding, ND 62.5 (62.3) ϩ6.9 (ϩ4.8) ND (ND) ϩ2.1 NS et al,9 120 s (WAIS-R)22 2002 Digit span–backward ND 6.7 (7.1) ϩ1.7 (ϩ0.4) ND (ND) ϩ1.3 NS (WAIS-R)22 Executive function ND 2.50 (2.51) −0.03 (−0.12) ND (ND) ϩ0.09 NS (Stroop)23 Verbal fluency–initial ND 29.3 (23.7) −0.1 (ϩ3.3) ND (ND) −3.4 NS letter24 Verbal ability–vocabulary24 ND 23.3 (21.9) −0.3 (−0.1) ND (ND) −0.1 NS Sommer 20 (10 wk) 4 (3) Dementia, with Poor/narrow Memory scale–logical ND 4.9 (4.6) 0.0 (ϩ1.7) ND (ND) −1.7 NS et al,16 normal memory subtest 2003 folate levels (WAIS-R)30 (77) Memory scale–associate ND 16.6 (11.0) −7.8 (ϩ0.3) ND (ND) −7.5 .08 ሻ learning subtest (WAIS-R)30 WAIS-R prorated verbal IQ ND 107.8 (110.0) −2.5 (ϩ12.5) ND (ND) ϩ10.0 NS (WAIS-R)30 Boston naming test40 ND 40.8 (42.3) ϩ1.2 (ϩ1.4) ND (ND) −0.2 NS Controlled oral word ND 29.5 (36.0) ϩ3.3 (−5.0) ND (ND) ϩ8.3 NS association test41 Speed/concentration– ND 233 (278) ϩ15 (−17) ND (ND) ϩ32 NS Trails A32 Speed/concentration– ND 373 (412) ϩ20 (−55) ND (ND) ϩ75 .08 Trails B32 Speed/concentration– ND 38.4 (32.7) −1.5 (−8.4) ND (ND) ϩ6.9 NS finger-tapping test42

Abbreviations: ND, no data; NS, nonsignificant; Stroop, Stroop Color-Word Test; WAIS-R, Wechsler Adult Intelligence Scale–Revised. *Higher score indicates better cognitive function. †Lower than 3 ng/mL (Ͻ6.8 nmol/L). ‡Part of the Randt memory test.43

§Seventy-five participants were randomized among vitamins B6,B12, folic acid, and placebo. ሻPlacebo better than B vitamin.

nificant improvement was reported worsening of cognitive function with The 3 trials had discordant re- 13 in 1 test, in which a statistically sig- vitamin B12 supplementation. There sults in 3 different populations. In nificant change compared with base- was no evidence across trials of dif- a trial of relatively low-dose folic acid line was found with treatment, but ferences in effect in tests of differ- supplementation (0.75 mg/d) in the net change failed to reach statis- ent cognitive domains. women with normal cognitive func- tical significance. tion,9 no significant effects were Seal et al12 directly evaluated the FOLIC ACID found across 5 measured tests, with effect of vitamin B12 oral supplemen- 3 showing a net improvement and tation in 2 intervention arms, one re- Three trials of moderate to poor 2 a net worsening. In a very small ceiving double the dose of the other, quality reported data on folic acid trial of folic acid, 20 mg/d, in 7 and compared these groups with pla- supplementation and the effect on people with dementia and normal cebo. No significant change was cognitive function or therapeutic baseline folate levels,16 no statisti- found in Mini-Mental State Exami- benefit in a total of 39 people who cally significant difference was found nation26 scores when the 3 groups received folic acid (Table 4).9,15,16 in the results of 8 cognitive func- were compared. Two trials were conducted among tion tests, although there was a trend There is large heterogeneity participants with dementia or cog- toward a net worsening of associ- among trials in terms of dose, ad- nitive impairment, and 1 among nor- ate learning and a trend toward a net ministration route, and duration of mal participants. The trials tested improvement in 1 concentration test treatment, and it would be difficult various doses of folic acid ranging (Trails B32). Overall, 5 tests found a to support any conclusion about a from 0.75 to 20 mg/d. In compari- net improvement and 3 a net wors- potential dose effect. Overall, half of son, the US RDA for folate is 0.4 to ening of cognitive function. In the the tests found a net improvement 1 mg/d. Trial durations ranged from third trial of cognitively impaired in- in cognitive function and half a net 5 to 10 weeks. dividuals with low baseline folate

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 26

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 5. Effect of Combination Interventions on Cognitive Function

B Vitamin Intervention Dose, mg/d Participants Population (Control) Change Net Change (Controls), Characteristic Quality/ Maximum Base

Source, y FA B6 B12 Duration No. (Mean Age, y) Applicability Test Score* Value Value P Value Value P Value McMahon 1.0 10.0 0.5 2 y 125 (124) Normal (73) Good/wide MMSE26 30 29.2 (29.2) ϩ0.1 ND (ND) −0.09 NS et al,44 (Oral) (ϩ0.2) 2006 Wechsler paragraph 50 22.5 (23.4) −3.8 ND (ND) −0.88 NS recall test22 (−2.7) Category word ND 54.2 (55.8) ϩ11.6 ND (ND) −0.86 NS fluency test, total (ϩ13.0) words in 3 minutes45 Rey auditory verbal 75 42.1 (42.6) ϩ1.8 ND (ND) −0.98 NS learning test trials (ϩ1.6) I-V45 Rey auditory verbal 15 7.8 (7.7) −0.5 ND (ND) −0.35 NS learning test trials (−0.2) VI45 Raven progressive 20 14.2 (14.2) −2.6 ND (ND) −0.31 NS matrices36 (−2.3) Controlled oral word ND 37.6 (39.3) ϩ2.5 ND (ND) ϩ0.31 NS association test41 (ϩ1.7) Reitan trail-making ND† 104.8 (100.7) ϩ9.6 ND (ND) ϩ1.08 .007 test, part B32 (−1.7) Stott 2.5 NG 0.5 3/12 85/73 Ischemic Good/ Telephone interview 39 25.5 (I&C) −0.5 NS (NS) −0.8 NS et al,8 (Oral) mo‡ (82/73)§ vascular narrow for cognitive (ϩ0.3) 2005 disease (74) status20 Letter-digit coding ND 19.7 (I&C) −0.85 NS (NS) −0.9 NS test21 (ϩ0.1) Lewerin 0.8 3.0 0.5 4 mo 115 (64) Normal (76) Fair/wide Digit span–forward 9 5.8 (5.9) ϩ0.2 .09 (NS) −0.1 NS et al,17 (Oral) (WAIS)46 (ϩ0.3) 2005 Digit 8 4.4 (4.6) ϩ0.25 .09 (NS) ϩ0.03 NS span–backward (ϩ0.22) (WAIS)46 Block design 42 18.5 (20.0) ϩ1.0 NS (NS) ϩ0.2 NS (WAIS)46 (ϩ0.8) Digit symbol 90 35.1 (38.0) ϩ0.9 NS (NS) −1.4 .09 ࿣ (WAIS)46 (ϩ2.3) Identical forms47 60 23.3 (24.8) ϩ0.1 NS (NS) −1.4 .04 ࿣ (ϩ1.5) Visual 14 6.9 (7.0) ϩ0.6 NS (NS) 0 NS reproduction38 (ϩ0.6) Synonyms47 30 22.5 (22.4) ϩ0.31 NS (NS) −1.0 .02 ࿣ (ϩ1.3) Thurstone picture 28 20.3 (21.1) ϩ1.7 NS (NS) −0.7 NS memory test48 (ϩ2.4) Figure 30 15.8 (16.8) ϩ1.5 NS (NS) ϩ0.9 NS classification48 (ϩ0.6) Eussen 0.4 NG 1 (Oral) 24 wk 66 (65) Normal to Fair/ Digit span–forward 16 7.5 (7.6) −0.1 NS (NS) −0.3 NS et al,29 cognitive moderate (WAIS)22 (ϩ0.2) 2006 impairment Digit 14 5.1 (4.7) −0.2 Ͻ.05 −0.8 .001 ࿣ (73) span–backward (ϩ0.6) (Ͻ.05) (WAIS)22 Executive function 12 4.7 (4.8) ϩ1.1 NS (NS) ϩ0.5 NS similarities (ϩ0.6) (WAIS)30 Speed–finger ND† 442 (409) −17 NS (NS) ϩ3NS tapping, ms31 (−20) Speed–trail-making ND† 76.9 (72.0) −7.1 NS (NS) −9.0 NS test, part A, s32 (ϩ1.9) 15-Word 75 30.1 (30.0) ϩ6.2 NS (NS) ϩ0.5 NS learning– (ϩ5.7) immediate recall33 15-Word 15 4.6 (5.1) ϩ1.5 NS (NS) ϩ0.5 NS learning–delayed (ϩ1.0) recall33 15-Word 30 26.6 (25.3) ϩ0.8 Ͻ.05 −0.9 .03 ࿣ learning– (ϩ1.7) (Ͻ.05) recognition33 Executive ND† 2.7 (2.9) ϩ0.4 NS (NS) ϩ0.5 NS function–trail- (−0.1) making test, part C/part A32 Executive ND† 2.2 (2.1) 0.0 NS (NS) −0.7 NS function–Stroop (ϩ0.7) test, part 3/part 234

(continued)

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 27

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 5. Effect of Combination Interventions on Cognitive Function (cont)

B Vitamin Intervention Dose, mg/d Participants Population (Control) Change Net Change (Controls), Characteristic Quality/ Maximum Base

Source, y FA B6 B12 Duration No. (Mean Age, y) Applicability Test Score* Value Value P Value Value P Value Eussen 0.4 NG 1 (Oral) 24 wk 66 (65) Normal to Fair/ Executive 24 15.2 ϩ1.5 NS ϩ0.5 NS et al,29 cognitive moderate function– (15.5) (ϩ1.0) (NS) 2006 impairment Raven36 (73) Word fluency, ND 17.6 −0.3 NS ϩ0.6 NS animals, No. (17.4) (−0.9) (NS) of nouns37 Word fluency, ND 15.5 ϩ0.5 Ͻ.05 −1.8 NS letters, No. (15.2) (ϩ2.3) (Ͻ.05) of nouns37 Clarke 2 NG 1 (Oral) 12 wk 128¶ Dementia Poor/ MMSE26 30 21 Not significantly altered by et al,18 (75) moderate (I&C) treatment 2003 ADAS49 70† 27 Not significantly altered by (I&C) treatment Shaw 15 NG 1# 12 wk 17¶ Severe Poor/narrow Dementia scale50 ND ND The means were unchanged in both et al,19 (Injection) dementia groups 1971 (81) Blessed ND ND Increment not statistically Information– significant Memory Concentration Test51

Abbreviations: ADAS, Alzheimer Disease Assessment Scale; B6, pyridoxine hydrochloride (vitamin B6,); B12, cyanocobalamin or hydroxycobalamin (vitamin B12); FA, folic acid; I&C, intervention and control groups not reported separately; MMSE, mini-mental state examination; ND, no data; NG, not given; NS, nonsignificant; WAIS, Wechsler Adult Intelligence Scale. *Higher score indicates better cognitive function unless otherwise noted. †Lower score indicates better cognitive function. ‡Intervention lasted 12 weeks; follow-up cognitive testing occurred at 1 year. §Number tested with telephone interview for cognitive status20/number tested with letter-digit coding test.21 ࿣Placebo better than B vitamin. ¶No data on how many participants in each arm. #Hydroxycobalamin, 1000 mg/d for the first week and then 100 mg/wk for 11 weeks thereafter.

levels,15 supplementation with folic tially higher than the US RDA for There were no dose-related re- acid, 15 mg/d, resulted in a net ben- each vitamin. Studies tested combi- sponses discussed in the trials. There efit in all 6 cognitive tests used, 4 of nation B vitamins in from 17 to 128 was no evidence across trials of dif- which were statistically significant. participants. Two studies used a ferences in effect on tests of differ- The study found that the cognitive combination of folic acid, vitamin B6, ent cognitive domains. 17,44 improvement after folic acid inter- and vitamin B12 ; the other 4 tri- vention was correlated in a linear als combined folic acid and vita- 8,18,19,29 COMMENT fashion with the low levels of folate min B12. One study used vi- at baseline. tamin B12 injections and oral folic Owing to the paucity of evi- acid,19 while the other trials used oral The limited evidence from the short- dence evaluating the effect of folic vitamin B12. One study lasted 2 duration randomized controlled tri- acid on the outcomes, any conclu- years44; in the others, the B vitamin als suggests that supplementation sions remain tentative. Neverthe- supplements were given for either 3 with vitamins B6 or B12 or folic acid less, the limited evidence may sug- or 4 months; though 1 trial tested among either elderly cognitively in- gest a benefit from folic acid cognitive function 9 months after tact individuals or those with de- supplementation for people with ending treatment.8 mentia or cognitive impairment does both cognitive impairment and se- For most of the cognitive tests, no not improve cognitive function as rum folate levels lower than 3 ng/mL effect was found with B vitamin treat- measured by a wide battery of tests. (Ͻ6.8 nmol/L). ment. In 2 studies,17,29 participants in The 1 exception to this may be folic the placebo arm performed better acid supplementation in people with COMBINED B VITAMINS than those receiving B vitamins in 5 both cognitive impairment and se- of the 25 tests. In contrast, McMahon rum folate levels lower than 3 ng/mL Six trials of varying quality re- et al44 found a significant improve- (Ͻ6.8 nmol/L). However, any con- ported data on the effect of comment in 1 of 8 cognitive tests with a clusions about the effect of folic acid bined B vitamin intervention in par- combination of all 3 B vitamins. The treatment are greatly limited be- ticipants with dementia, normal remaining 3 trials of combination folic cause there are only 3 trials with 39 cognitive function, or ischemic vas- acid and vitamin B12 found no signifi- people who received folic acid. Fur- cular disease (Table 5).8,17-19,29,44 All cant effects on cognitive testing, al- ther clarification may be available trials used different daily doses of though 2 did not report quantitative soon because a large, 3-year trial of various B vitamins; all substan- data.8,18,19 folic acid in elderly people with hy-

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 28

Downloaded From: https://jamanetwork.com/ on 10/01/2021 perhomocysteinemia has been pre- are necessary to address whether B Correspondence: Ethan M. Balk, sented at an international confer- vitamin supplementation is effec- MD, MPH, Institute for Clinical Re- ence.52 The abstract, which has not tive to slow cognitive decline or im- search and Health Policy Studies, yet been peer reviewed, preliminar- prove cognitive function. No peer- Tufts-New England Medical Cen- ily reported a net improvement in reviewed trial has addressed the ter, 750 Washington St, NEMC No. global cognitive function including question of whether these B vita- 63, Boston, MA 02111 (ebalk memory, sernsorimotor speed, and mins can prevent or delay clinical de- @tufts-nemc.org). informational processing speed. mentia or can affect clinical symp- Author Contributions: Dr Balk had Notably, in a minority of studies toms of dementia. However, further full access to all of the data in the of both vitamin B12 and combina- standardization is clearly needed study and takes responsibility for the tions of B vitamins that include vi- both in the field of vitamin re- integrity of the data and the accu- tamin B12 compared with placebo, search and for the assessment of cog- racy of the data analysis. Study con- the B vitamin treatment resulted in nitive function. Future studies cept and design: Balk, Chung, Lau, worse performance on a small num- should use only well-established di- and Rosenberg. Acquisition of data: ber of cognitive tests. The clinical agnostic criteria for neurocognitive Balk, Raman, and Chung. Analysis significance of this effect, though, is disorders and only measures of cog- and interpretation of data: Balk, Ra- questionable, in that those receiv- nitive function that have been veri- man, Tatsioni, and Rosenberg. Draft- ing placebo had statistically signifi- fied and accepted by the neurocog- ing of the manuscript: Balk. Critical cant improvements from baseline nitive research community. Studies revision of the manuscript for impor- among these tests. that use nonstandard diagnostic defi- tant intellectual content: Balk, Ra- Overall, the data are sparse, and nitions or neurocognitive tests are man, Tatsioni, Chung, Lau, and firm conclusions are not possible. Im- of lesser value to clinicians, policy Rosenberg. Statistical analysis: Balk portantly, no trial evaluated the risk makers, and other researchers. Re- and Chung. Obtained funding: Balk ofdevelopingdementia,clinicalsymp- searchers should also preferen- and Lau. Administrative, technical, toms, neurological imaging tests, or tially use cognitive function tests that and material support: Raman, Chung, clinical outcomes beyond ability to adequately differentiate between dif- and Lau. Study supervision: Balk and perform cognitive function tests. ferent cognitive domains and should Lau. About 50 different tests or subtests evaluate how B vitamin supplemen- Financial Disclosure: None re- were used across the trials, measur- tation differentially affects differ- ported. ing different or overlapping domains ent cognitive domains. In addition, Funding/Support: This evidence re- of cognitive function; thus, compari- imaging studies such as magnetic port was prepared by the Tufts- sons across studies were difficult. In resonance imaging and positron- New England Medical Center Evi- addition, studies did not explicitly re- emission tomography may prove to dence-based Practice Center under port the clinical importance of the be of value to assess how B vitamin contract to the Agency for Health- changes in cognitive function test re- supplementation affects cognitive care Research and Quality (con- sults.Consequently,clinicalinferences function. tract No. 290-02-0023), Rockville, based on the reported scores alone In conclusion, the few available Md. Funding was provided by the would be problematic. randomized controlled trials of vi- Office of Dietary Supplements and Only 1 trial compared different tamins B6, and B12 and folic acid the National Center for Comple- doses of vitamin B12, and only 2 di- supplementation, alone or in com- mentary and Alternative Medicine, rectly compared the different B vi- bination, do not provide adequate National Institutes of Health. tamins.8,9 No trial directly com- evidence for a beneficial effect of pared populations, routes of supplementation on cognitive func- REFERENCES administration of vitamin B12, or dif- tion testing in people with either ferent durations. In addition, the normal or impaired cognitive func- small number of available trials lim- tion, although 1 very small trial 1. Ferri CP, Prince M, Brayne C, et al. Global preva- lence of dementia: a Delphi consensus study. ited our ability to assess the impact found an improvement with folic Lancet. 2005;366:2112-2117. of these factors. The quality of the acid supplementation in patients 2. Herrmann N. Cognitive pharmacotherapy of Alz- trials was often compromised by in- with low baseline folate levels. How- heimer’s disease and other dementias. Can J complete reporting of methodol- ever, the poor quality and the small Psychiatry. 2002;47:715-722. 3. Wardlaw GM, Kessel M. The Water-Soluble Vita- ogy and results, lack of blinding, number of trials and participants mins: Perspectives in Nutrition. 5th ed. New York, small sample size, short duration, preclude a firm conclusion. Addi- NY: McGraw-Hill; 2002:364-417. and other factors. Two of the trials tional trials are needed to deter- 4. Riggs KM, Spiro A III, Tucker K, Rush D. Rela- also implicitly tested the effect of mine which populations, if any, tions of vitamin B-12, vitamin B-6, folate, and ho- short-term intervention (1 or 3 would benefit from B vitamin mocysteine to cognitive performance in the Nor- mative Aging Study. Am J Clin Nutr. 1996;63: months) on longer-term cognitive supplementation and to determine 306-314. 8,11 function (3 or 12 months). Inter- whether B vitamin supplementa- 5. Wang HX, Wahlin A, Basun H, Fastbom J, Win- pretation of the clinical value of this tion would affect clinical signs and blad B, Fratiglioni L. Vitamin B(12) and folate in type of trial is unclear. symptoms of dementia. relation to the development of Alzheimer’s disease. Neurology. 2001;56:1188-1194. Future well-performed, well- 6. Balk E, Chung M, Raman G, et al. B Vitamins and analyzed, sufficiently powered, long- Accepted for Publication: Septem- Berries and Age-Related Neurodegenerative Dis- term, randomized controlled trials ber 15, 2006. orders: Evidence Report/Technology Assess-

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 29

Downloaded From: https://jamanetwork.com/ on 10/01/2021 ment No. 134. Rockville, Md: Agency for Health- and homocysteine in people at high risk of B): A Manual. Lisse, the Netherlands: Swets & care Research and Quality; April 2006. dementia. J Intern Med. 2003;254:67-75. Zeitlinger; 1986. 7. National Kidney Foundation. K/DOQI clinical prac- 19. Shaw DM, Macsweeney DA, Johnson AL, et al. 34. Houx PJ, Jolles J, Vreeling FW. Stroop interfer- tice guidelines for chronic kidney disease: evalu- Folate and amine metabolites in senile dementia: ence: aging effects assessed with the Stroop Color- ation, classification, and stratification. AmJKid- a combined trial and biochemical study. Psychol Word Test. Exp Aging Res. 1993;19:209-224. ney Dis. 2002;39(suppl 1):S1-S266. Med. 1971;1:166-171. 35. Visser RH. Manual of the Complex Figure Test. 8. Stott DJ, MacIntosh G, Lowe GD, et al. Random- 20. Gallo JJ, Breitner JC. Alzheimer’s disease in the Lisse, the Netherlands: Swets & Zeitlinger; 1985. ized controlled trial of homocysteine-lowering vi- NAS-NRC Registry of aging twin veterans, IV: per- 36. Raven J. Guide to Using the Colored Progressive tamin treatment in elderly patients with vascular formance characteristics of a two-stage tele- Matrices. London, England: HK Lewis; 1965. disease. Am J Clin Nutr. 2005;82:1320-1326. phone screening procedure for Alzheimer’s 37. Luteijn FvdPF. Manual Groningen Intelligence Test. 9. Bryan J, Calvaresi E, Hughes D. Short-term folate, dementia. Psychol Med. 1995;25:1211-1219. Lisse, the Netherlands: Swets & Zeitlinger; 1983. vitamin B-12 or vitamin B-6 supplementation 21. Houx PJ, Shepherd J, Blauw GJ, et al. Testing cog- 38. Wechsler D. Standardized memory scale for clini- slightly affects memory performance but not mood nitive function in elderly populations: the PROSPER cal use. J Psychol. 1945;19:87. in women of various ages. J Nutr. 2002;132: study: PROspective Study of Pravastatin in the Eld- 39. Cameron DE. Impairment in the retention phase 1345-1356. erly at Risk. J Neurol Neurosurg Psychiatry. 2002; of remembering. Psychiatr Q. 1943;17:395. 10. Deijen JB, van der Beek EJ, Orlebeke JF, van den 73:385-389. 40. Kaplan E, Goodglass H, Weintraub S. The Bos- Berg H. Vitamin B-6 supplementation in elderly 22. Wechsler D. Wechsler Adult Intelligence Scale. 3rd ton Naming Test. Philadelphia, Pa: Lea & Fe- men: effects on mood, memory, performance and ed. San Antonio, Tex: Psychological Corp; 1997. biger; 1983. mental effort. Psychopharmacology (Berl). 1992; 23. Dodrill CB. A neuropsychological battery for 41. Benton AL, Hamsher K, Sivan AB. Multilingual 109:489-496. epilepsy. Epilepsia. 1978;19:611-623. Aphasia Examination. Iowa City, Iowa: AJA As- 11. Hvas AM, Juul S, Lauritzen L, Nexo E, Ellegaard 24. Benton AL, Hamsher D. Multilingual Aphasia sociates; 1994. J. No effect of vitamin B-12 treatment on cogni- Examination. Iowa City: University of Iowa; 1989. 42. Reitan RM. Clinical Neuropsychology: Current Sta- tive function and depression: a randomized pla- 25. Emmen HH, Hogendijk EMG, Hooisma J, Orle- tus and Applications. New York, NY: Hemi- cebo controlled study. J Affect Disord. 2004; beke JF, Uijtdehaage SHJ. Adaption of Two Stan- sphere; 1974. 81:269-273. dardized International Test Batteries for Use in the 43. Randt CT, Brown ER. Randt Memory Test. Bay- 12. Seal EC, Metz J, Flicker L, Melny J. A random- Netherlands for Detection of Exposure to Neuro- port, NY: Life Science Associates; 1983. ized, double-blind, placebo-controlled study of oral toxic Compounds: Internal Report. Rijswijk, the 44. McMahon JA, Green TJ, Skeaff CM, Knight RG, vitamin B12 supplementation in older patients with Netherlands: Medisch Biologisch Laboratorium Mann JI, Williams SM. A controlled trial of ho- subnormal or borderline serum vitamin B12 TNO; 1988. mocysteine lowering and cognitive performance. concentrations. J Am Geriatr Soc. 2002;50:146- 26. Folstein MF, Folstein SE, McHugh PR. “Mini- N Engl J Med. 2006;354:2764-2772. 151. mental state”: a practical method for grading the 45. Lezak MD. Neuropsychological Assessment. 3rd 13. Kwok T, Tang C, Woo J, Lai WK, Law LK, Pang cognitive state of patients for the clinician. JPsy- ed. New York, NY: Oxford University Press; 1995. CP. Randomized trial of the effect of supplemen- chiatr Res. 1975;12:189-198. 46. Wechsler D. The Measurement and Appraisal of tation on the cognitive function of older people with 27. Roth M, Tym E, Mountjoy CQ, et al. CAMDEX: a Adult Intelligence. Baltimore, Md: Williams & subnormal cobalamin levels. Int J Geriatr standardised instrument for the diagnosis of men- Wilkins; 1958. Psychiatry. 1998;13:611-616. tal disorder in the elderly with special reference 47. Dureman I, Salde H. Psychometric and Experi- 14. Kral VA, Solyom L, Enesco H, Ledwidge B. Rela- to the early detection of dementia. Br J Psychiatry. mental Psychological Methods for Clinical Appli- tionship of vitamin B12 and folic acid to memory 1986;149:698-709. cation [in Swedish]. Stockholm, Sweden: Alm- function. Biol Psychiatry. 1970;2:19-26. 28. Nielsen H, Lolk A, Krag-Sorensen P. Normative quist & Wiksell; 1959. 15. Fioravanti M, Ferrario E, Massaia M, et al. Low folate data for eight neuropsychological tests, gath- 48. Thurstone L. Primary Mental Abilities: Psycho- levels in the cognitive decline of elderly patients ered from a random sample of Danes aged 64-83 metric Monographs. Chicago, Ill: University of Chi- and the efficacy of folate as a treatment for im- years. Nord Psykol. 1995;47:241-255. cago Press; 1938. proving memory deficits. Arch Gerontol Geriatr. 29. Eussen SJ, de Groot LC, Joosten LW, et al. Effect 49. Rosen WG, Mohs RC, Davis KL. A new rating scale 1997;26:1-13. of oral vitamin B-12 with or without folic acid on for Alzheimer’s disease. Am J Psychiatry. 1984; 16. Sommer BR, Hoff AL, Costa M. Folic acid supple- cognitive function in older people with mild vita- 141:1356-1364. mentation in dementia: a preliminary report. J Geri- min B-12 deficiency: a randomized, placebo- 50. Kendrick DC. A cross-validation study of the use atr Psychiatry Neurol. 2003;16:156-159. controlled trial. Am J Clin Nutr. 2006;84:361- of the SLT and DCT in screening for diffuse brain 17. Lewerin C, Matousek M, Steen G, Johansson B, 370. pathology in elderly subjects. Br J Med Psychol. Steen B, Nilsson-Ehle H. Significant correlations 30. Wechsler D. Wechsler Adult Intelligence Scale– 1967;40:173-178. of plasma homocysteine and serum methylma- Revised Manual. New York, NY: Psychological 51. Blessed G, Tomlinson BE, Roth M. The associa- lonic acid with movement and cognitive perfor- Corp, Harcourt Brace & Jovanovich; 1982. tion between quantitative measures of dementia mance in elderly subjects but no improvement 31. Houx P. Cognitive Aging and Health-Related and of senile change in the cerebral grey matter from short-term vitamin therapy: a placebo- Factors. Maastricht, the Netherlands: Maastricht of elderly subjects. Br J Psychiatry. 1968;114: controlled randomized study. Am J Clin Nutr. 2005; University; 1991. 797-811. 81:1155-1162. 32. Reitan RM. Trail Making Test: Manual for Admin- 52. Durga J. Effect of 3-year folic acid supplementa- 18. Clarke R, Harrison G, Richards S; Vital Trial Col- istration and Scoring. Tucson, Ariz: Reitan Neu- tion on cognitive function in older adults: a ran- laborative Group. Effect of vitamins and aspirin ropsychological Laboratory; 1992. domized double blind controlled trial [abstract]. on markers of platelet activation, oxidative stress 33. Saan RJ, Deelman BJ. New 15-Words Test (A and J Nutr Health Aging. 2006;10:204-214.

(REPRINTED) ARCH INTERN MED/ VOL 167, JAN 8, 2007 WWW.ARCHINTERNMED.COM 30

Downloaded From: https://jamanetwork.com/ on 10/01/2021