ARTICLES

Childhood cognitive ability and risk of

late-onset Alzheimer and vascular dementia

Brian McGurn, MRCP ABSTRACT

Ian J. Deary, PhD Background: Risk factors for dementia can be divided into those that lead to pathologic insults
John M. Starr, FRCPEd and those that render the brain more vulnerable to such insults. Dementia prevention strategies
need to be informed as to whether to prioritize reducing risk of pathology or to reduce the vulner-
ability of the brain to pathologic insult. Lower premorbid cognitive ability is associated with both
increased vascular risk and reduced cognitive reserve, a measure of brain vulnerability. We inves-
Address correspondence and tigated differential effects of premorbid cognitive ability on dementia subtypes to clarify which
reprint requests to Professor John
M. Starr, Royal Victoria Hospital,
causal pathway was predominant.
Craigleith Road, Edinburgh EH4 Methods: A total of 297 cases of late onset dementia were identified from local case registers,
2DN, UK
[email protected] born in 1921, and thus may have participated in the Scottish Mental Survey of 1932
(SMS1932). A total of 183 had mental ability scores identified in the SMS1932. A total of 173 of
these were matched by birth records to one set of controls (controls1) by date of birth, sex, and
district of birth registration. A further set of controls (controls2) was generated additionally
matched on paternal occupation.
Results: Vascular dementia cases had significantly lower premorbid cognitive ability (OR 0.62,
95% CI 0.41– 0.94, for every 10-point increase [0.7 SD] in mental ability score vs controls2), but
there was no significant difference in the premorbid cognitive ability of Alzheimer disease cases
(OR 1.02, 95% CI 0.82–1.28) vs controls2.
Conclusion: Lower premorbid cognitive ability is a risk factor for vascular dementia, but not Alz-
heimer disease. This suggests its effect acts mainly through vascular pathology rather than brain
vulnerability. Neurology® 2008;71:1051–1056

GLOSSARY
AD ⫽ Alzheimer disease; CBF ⫽ cerebral blood flow; LMTC ⫽ Lothian Memory Treatment Centre; MHT ⫽ Moray House Test;
REH ⫽ Royal Edinburgh Hospital; SMS1932 ⫽ Scottish Mental Survey of 1932; VaD ⫽ vascular dementia.

Dementia is predicted to be one of the greatest worldwide disease burdens in the 21st century
with one new case occurring every 7 seconds.1 Moreover, in the United Kingdom at least, costs
of dementia are predicted to far outstrip those that would be predicted due to rising prevalence
associated with demographic change.2 Relatively small changes in prevalence are likely to result
in substantial cost savings2 and could be achieved by modest delays to the onset of dementia.3
Alzheimer disease (AD), the commonest cause of dementia, has an insidious onset with a
preclinical phase preceding diagnosis by many years.4 Neuropathologic features of AD, plaques
and tangles, are thus typically present for some time before the diagnostic threshold for demen-
tia is reached. Concomitant vascular lesions increase the likelihood of a clinical diagnosis of
dementia,5 though both neuropathologic features of AD and vascular lesions are common in
the brains of people without dementia.6 These findings indicate limitations with using neuro-
pathologic standards for clinical diagnostic criteria for both AD and vascular dementia (VaD)
but, at present, alternative gold standards are unavailable. The hypothesis that vascular lesions
Editorial, page 1046
add to the pathologic load that pushes someone over the dementia threshold is supported by
See also pages 1057
and 1065
e-Pub ahead of print on June 25, 2008, at www.neurology.org.
From the Geriatric Medicine Unit (B.M., J.M.S.) and Psychology (I.J.D.), University of Edinburgh, UK.
Supported by a clinical research training fellowship (B.M.). Ian J. Deary is the recipient of a Royal Society-Wolfson Research Merit Award.
Disclosure: The authors report no disclosures.

Copyright © 2008 by AAN Enterprises, Inc. 1051

 data from studies that show vascular risk fac- duce cognitive decline as much as expected.17
tors in midlife are associated with increased However, a priori vascular risk factors might be
AD incidence in old age.7,8 One prevention expected to be more influential on VaD than
strategy, therefore, takes a life course perspec- AD incidence, so investigating differential ef-
tive, aiming to reduce vascular risk in middle fects of premorbid cognitive ability on different
age. A complementary approach does not fo- dementia subtypes may help clarify where pre-
cus on the pathologic load, but on characteris- vention effort is best directed.
tics of the individual that might increase the
METHODS Measure of premorbid cognitive ability.
load needed to cross the diagnostic threshold.
This is provided by the Scottish Mental Survey of 1932
This strategy usually invokes the concept of (SMS1932).18,19 On June 1, 1932, almost every child at school in
cognitive reserve9,10 that posits that aspects of Scotland who was born in the calendar year 1921 took a group-
the brain’s structure and function serve to administered mental ability test. In total 87,498 children
(44,210 boys and 43,288 girls) took the test, representing about
buffer the effects of neuropathology.11 Hence 95% of the relevant population. The test used was a version of
increasing cognitive reserve could delay de- the Moray House Test No. 12 (MHT), which consists of 71
mentia onset for the same pathologic load. items in categories including following directions, same-
Many factors are thought to contribute to opposites, analogies, reasoning, arithmetical, spatial, mixed sen-
tences, and proverbs. The maximum possible score is 76. To
cognitive reserve, but a key index that cap- assess the criterion validity of the MHT, a random subset of the
tures the effects of many of these is premorbid birth year (500 boys and 500 girls) were given the Stanford-
cognitive ability.11 Binet test, administered individually. The correlation between
the Stanford-Binet test and MHT was 0.81 for boys and 0.78 for
Premorbid cognitive ability is not only a key
girls, thus validating the MHT as a test of general mental ability.
index of cognitive reserve, but predicts surviv- The results of the SMS1932 were recorded in handwritten led-
al,12 cardiovascular disease,13 and midlife gers based on geographic area. These contain the pupils’ names,
blood pressure.14 Theoretically, an association schools, and MHT scores. These ledgers are stored in the Scot-
tish Council for Research in Education archive at the University
between lower premorbid cognitive ability and
of Glasgow, Scotland. The data had been transcribed onto a
dementia would be expected to act not only by computerized database. This database was made available by the
reduced cognitive reserve, but also by greater Scottish Centre for Research in Education (SCRE) to the re-
pathologic load, especially with respect to vascu- search team involved in follow-up studies of the SMS1932.

lar lesions. No such association has been de- Definition of cases. Selection of cases and controls followed
tected for early onset dementia, but lower IQ at the methods of an earlier study of early onset dementia.15 Like
one subgroup in the earlier study, all cases and controls in this
age 11 has been associated with increased risk of study had an MHT score available on the SMS1932 database
late-onset dementia.15 However, this study com- and thus were all born in 1921. This sample was thus drawn
prised only 50 late-onset cases, so was inade- from the same birth cohort, but with cases developing AD after
rather than before 65 years of age. Hence there was no overlap in
quately powered to determine effects of
cases between the earlier study and this one. A case was defined
childhood IQ on AD and VaD separately. The as someone who had had an International Classification of Dis-
Nun Study also found an association between eases (ICD)-1020 diagnosis of dementia made after the age of 65
linguistic ability at about 22 years of age and years. A total of 297 cases were identified from dementia case
registers in Edinburgh maintained by the main psychiatric hos-
late-onset dementia, but this was based on just
pital, the Royal Edinburgh Hospital (REH), and a tertiary mem-
39 neuropathology cases, only 59% of whom ory clinic, Lothian Memory Treatment Centre (LMTC).
fulfilled clinical diagnostic criteria for demen- Referrals were made to the LMTC from the whole of the region
tia.16 Intuitively, a causal chain linking lower of Lothian which has Edinburgh as its major city. The latest date
for case identification was August 2003 when cases would have
premorbid cognitive ability to midlife vascular been 82 years old. The expected prevalence of dementia in the
risk factors and thence to dementia in old age population from which cases were drawn21,22 was estimated as
might be hypothesized. But demonstrating an 392 cases. Hence ascertainment was 75.8%, but this represents
an upper limit because there may have been a small number of
association between midlife vascular factors and
cases who had died since being entered on the case registers in-
late-onset dementia may be misleading because flating this figure. LMTC AD cases had been classified according
vascular risk is associated with lower premorbid to NINCDS diagnostic criteria.23 Its diagnostic criteria for VaD
cognitive ability and thus with reduced cogni- had varied between National Institute of Neurological Disorders
and Stroke–Association Internationale pour la Rechercheet
tive reserve. These bidirectional relationships be-
l’Enseignement en Neurosciences24 and ICD-10. In view of this
tween cognition and vascular risk may be one all cases were reviewed and classified according to ICD-10 diag-
reason why lowering blood pressure fails to re- nostic guidelines as AD or VaD. The ICD-10 criteria were op-

1052 Neurology 71 September 30, 2008

 childhood IQ and paternal social class that may have affected the
Table 1 Demographic data and mean MHT scores in cases in different
early onset study.15 Thus, eventually, there were two control
diagnostic groups
groups. Control group 1 was matched on age, sex, and district of
Type of dementia birth registration. Control group 2 was matched on age, sex,
district of birth registration, and father’s occupation. To increase
AD VaD/mixed Dementia (unspecified) power, each control group contained two individuals as con-
No. 86.0 40.0 44.0 trols—the person closest before and closest after the case on the
Age at SMS, y 10.9 (0.29) 10.9 (0.27) 10.9 (0.27)
register of births who met the matching criteria. Not infre-
quently those in control group 1 were the same as those in con-
Female sex, n (%) 58.0 (67) 22.0 (55) 22.0 (50)
trol group 2; that is, the controls in group 1 often shared paternal
Born Lothian, n (%) 63.0 (73) 31.0 (78) 33.0 (75) occupation with the case. Variables available that were not used
Maternal age, y 30.7 (7.0) 30.2 (5.7) 30.6 (6.5) for matching were paternal age at subject’s birth, maternal age at
Paternal age, y 33.7 (8.2) 32.0 (6.2) 33.0 (7.1)
subject’s birth, length of time parents had been married at sub-
ject’s birth, and age of the subject at SMS1932.
Years parents married 3.9 (1.2–10.5) 5.0 (1.2–10.0) 3.5 (1.3–9.5)
at birth of subject,
median (interquartile range)
Statistical analysis. After initial data description, compari-
sons between cases and controls were made with t tests and then
Paternal occupational category, n (%)
OR with 95% CI estimated using conditional logistic regression.
I 2.0 (2) 0.0 3.0 (7) Conditional logistic regression represents an extreme form of
II 7.0 (8) 4.0 (11) 4.0 (9) stratification to control for confounding exposures where each
III 50.0 (59) 25.0 (66) 24.0 (55)
case is matched to specific controls.

IV 16.0 (19) 6.0 (18) 8.0 (18)

RESULTS Eighty-six (49.7%) cases had AD, 32
V 10.0 (12) 3.0 (18) 5.0 (11)
(18.5%) VaD, 8 (4.6%) mixed AD/VaD, 44
MHT score 40.0 (14.5) 35.0 (13.1) 38.7 (15.6)
(25.4%) unspecified dementia, 2 (1.2%) PD, and
MHT ⫽ Moray House Test; AD ⫽ Alzheimer disease; VaD ⫽ vascular dementia; SMS ⫽ Scot-
one (0.6%) as “other neurodegenerative.” Demo-
tish Mental Survey. graphic data were similar for all main dementia sub-
types (table 1). Maternal age at birth was similar for
erationalized to make a diagnosis of VaD with an emphasis on
AD cases (mean 30.7 years, SD [SD] 7.0), VaD cases
dementia with features of 1) sudden onset of symptoms and
stepwise deterioration, 2) important vascular risk factors, focal
(30.2 years, SD 5.7), and unspecified dementia (30.6
neurologic signs, and supporting neuroimaging and neuropsy- years, SD 6.5), as was paternal age at birth (AD 33.7
chological assessment, or 3) a temporally appropriate clinical his- years, SD 8.2; VaD 32.0 years, SD 6.2; unspecified
tory of stroke or TIA. Some cases had features of both AD and dementia 33.0 years, SD 7.1). AD cases’ parents had
VaD. ICD-10 does not have a separate coding for mixed AD/ been married for a median 3.9 years (interquartile
VaD, but states, “In a certain proportion of cases, the features of
range 1.2–10.5 years) at the case’s birth, VaD cases’
AD and vascular dementia may both be present. In such cases,
double diagnosis (and coding) should be made. When the vascu- parents for 5.0 (1.2–10.0) years, and unspecified de-
lar dementia precedes the AD, it may be impossible to diagnose mentia parents for 3.5 (1.3–9.5) years. There was a
the latter on clinical grounds.”20 We classified cases that satisfied trend for VaD cases to have lower MHT scores than
both AD and VaD criteria as “mixed” and those that did not AD cases (t ⫽ 1.87, p ⫽ 0.06). Among cases, a sig-
fulfill either AD or VaD ICD-10 diagnostic criteria as “unspeci- nificant migration effect was found. Cases born out-
fied dementia.”
side Lothian (n ⫽ 43) had higher MHT scores than
Definition of controls. Controls used were population con- controls (44.6 vs 35.6 control group 1 and 35.0 con-
trols obtained from the national birth register held at Register trol group 2) and cases born in Lothian (44.6 vs 36.2,
House, Edinburgh. Therefore cases and controls had to have
n ⫽ 130, F ⫽ 11.3, p ⬍ 0.001). This was in the
personal details matched to that of individuals on both the
SMS1932 database (SMS1932) and the national register for context of the trend for people born in Lothian al-
births, deaths, and marriages. Of the 297 cases with late onset ready to have higher MHT scores than those born
dementia there were 44 females whose maiden name was not elsewhere in Scotland.19 Since we were unable to
known; therefore these cases were not able to be matched to the identify suitable controls who also migrated into
SMS. A further 65 cases could not be matched to the SMS; a
Lothian, cases who migrated into Lothian were ex-
high proportion of these cases are likely to have migrated to
Scotland after the age of 11, and so would not have taken part in
cluded from further analysis.
the SMS1932. A total of 183 cases were matched to the Comparison of MHT scores by dementia type
SMS1932. Of these, seven were not matched to the birth regis- showed that VaD cases had significantly lower scores
ter, probably indicating migration to Scotland before age 11 than both control group 1 (p ⫽ 0.02) and control
years. Three were adopted, and so were excluded as no data were group 2 (p ⫽ 0.01), but there was no significant dif-
known about paternal occupation required for matching con-
ference for AD cases (table 2). Conditional logistic
trols. This left 173 cases with which controls could be matched.
Each dementia case was matched individually in two ways to regression confirmed that higher MHT score age 11
population controls from the birth register. This was to try to was associated with significantly lower odds of VaD
avoid potential overmatching due to the correlation between (OR 0.62, 95% CI 0.41– 0.94 for a 10-point in-

Neurology 71 September 30, 2008 1053

 have lower MHT scores than control group 1 (mean
Table 2 Mean MHT score in cases and controls
in people born in Lothian
35.4 vs 40.7, p ⫽ 0.08) but not control group 2 (p ⫽
0.84).
Mean MHT p
Type of dementia No. (SD) t Value
DISCUSSION In this sample, lower premorbid cog-
AD
nitive ability increased risk of VaD but not AD. This
Case 63 37.7 (14.0)
association was independent of early childhood social
Control group 1 126 38.7 (15.0) ⫺0.5 0.64
status and the environmental factors captured here.
Control group 2 122 37.6 (14.9) 0.1 0.95 The data support the hypothesis that lower premor-
VaD bid cognitive ability acts as a risk factor for dementia,
Case 31 34.0 (13.5) through pathways of vascular risk rather than more
Control group 1 62 41.4 (14.5) ⫺2.4 0.02 generic cognitive reserve. If premorbid cognitive abil-
Control group 2 62 40.9 (11.8) ⫺2.5 0.01 ity was acting primarily as an index of cognitive re-
Unspecified
serve, a similar effect on AD and VaD would have
dementia been expected. By contrast, if the effect was primarily
Case 36 35.4 (16.0) through midlife vascular risk, low premorbid cogni-
Control group 1 72 40.7 (13.9) ⫺1.8 0.08 tive ability would, as reported here, have a greater
Control group 2 70 36.1 (14.6) ⫺0.2 0.84 impact on VaD. What has hitherto been considered
as a direct measure of brain structure and function25
MHT ⫽ Moray House Test; AD ⫽ Alzheimer disease; VaD ⫽
may, instead, be mediated by vascular disease. This
vascular dementia.
would be consistent with neuropathologic findings
crease in MHT score [0.7 standard deviations] com- from the Nun study where vascular lesions made a
pared with control group 2, table 3): no other major contribution to AD diagnosis.16 The same may
independent variable (paternal age at subject’s birth, be applicable to educational attainment, which is
maternal age at subject’s birth, length of time parents closely related to premorbid cognitive ability,26 and is
had been married at subject’s birth, age of subject at also associated with dementia risk.27 Education cor-
SMS) had any significant effect in the model. Results relates with the extent of the parietotemporal re-
for unspecified dementia cases were intermediate be- gional cerebral blood flow (CBF) deficit in AD and
tween AD and VaD. There was a trend for cases to this has been put forward as evidence of cognitive
reserve.28 However, such CBF deficits might also re-
Table 3 Relationship between MHT score and
sult from subtle vascular changes. The Framingham
dementia subtypes study demonstrated an association of education with
non-AD dementia, but not with AD.29 Similarly, a
Type of dementia OR (95% CI) p Value
study in Stockholm found an association between
AD
lower educational attainment and unspecified de-
Control group 1* 0.97 (0.79–1.19) 0.76 mentia but not AD.30 It might be argued that vascu-
Control group 2† 1.02 (0.82–1.28) 0.86 lar load is, itself, a component of cognitive reserve,
VaD but this would push the concept beyond the point of
Control group 1‡ 0.68 (0.50–0.94) 0.021 absurdity in the case of a vascular disease like VaD.
Control group 2§ 0.62 (0.41–0.94) 0.023 In this study, we identify a risk factor for VaD as
Unspecified dementia early as 11 years, well before completion of formal
Control group 1¶ 0.76 (0.56–1.04) 0.085
education, emphasizing the lifelong nature of risk
Control group 2

1.00 (0.76–1.33) 0.99
and that prevention strategies could be implemented
from childhood onwards. The unique availability of
*A total of 63 cases and 119 controls matched on sex, birth childhood cognitive data for a cohort now at high
registration district, and age at SMS.
risk of dementia together with data that show associ-
†A total of 61 cases and 119 controls matched on sex, birth
registration district, and paternal occupational category. ations between childhood cognitive ability and
‡A total of 31 cases and 60 controls matched on sex, birth midlife vascular disease independent of socioeco-
registration district, and age at SMS. nomic status and smoking13,14 allow causal pathways
§
A total of 31 cases and 60 controls matched on sex, birth
registration district, and paternal occupational category.
that stretch throughout the life course to be identi-
¶
A total of 33 cases and 63 controls matched on sex, birth fied. These data favor vascular risk reduction as a
registration district, and age at SMS. priority.


A total of 32 cases and 61 controls matched on sex, birth
Case representativeness was limited by use of
registration district, and paternal occupational category.
MHT ⫽ Moray House Test; AD ⫽ Alzheimer disease; VaD ⫽ case registers rather than independent identifica-
vascular dementia; SMS ⫽ Scottish Mental Survey. tion in the community. This also resulted in a

1054 Neurology 71 September 30, 2008

smaller sample size than would have been obtained ACKNOWLEDGMENT
by community screening: limited numbers may The authors thank Dr R. Lee for help with statistical analyses.

have meant that we missed a true difference in

Received September 10, 2007. Accepted in final form December 5,
MHT score between AD cases and controls be-
2007.
cause of inadequate power to detect an effect size
⬍0.5 standard deviations. Ascertainment is likely
to have been biased toward those seeking medical REFERENCES
help. If this were the case, we would have expected 1. Ferri CP, Prince M, Brayne C, et al. Global prevalence of
MHT score among people identified as having de- dementia: a Delphi consensus study. Lancet 2005;366:
2112–2117.
mentia to be higher than the population average.
2. McNamee P, Bond J, Buck D. Costs of dementia in En-
The mean MHT scores of cases born in Lothian of gland and Wales in the 21st century. Br J Psychiatr 2001;
36.2 was identical to the mean for the population 179:261–266.
of Lothian as a whole, slightly lower than the mean 3. Jorm AF, Dear KBG, Burgess NM. Projections of future
for the population of Edinburgh (37.3), but numbers of dementia cases in Australia with and without
higher than the Scottish mean of 34.5 (SD 15.5).18 prevention. Aust NZ J Psychiatry 2005;39:959–963.
It would be expected to be higher than the mean 4. Elias MF, Beiser A, Wolf PA, Au R, White RF,
D’Agostino RB. The preclinical phase of Alzheimer dis-
for all those scored in 1932 because attrition af-
ease: a 22-year prospective study of the Framingham co-
fects people with lower childhood IQs to a greater hort. Arch Neurol 2000;57:808–813.
extent.12 Controls were chosen by matching at 5. Snowdon DA, Greiner LH, Mortimer JA, Riley KP,
birth, but were known to have survived to 11 years Greiner PA, Markesbery WR. Brain infarction and the
of age as they participated in the SMS1932. VaD clinical expression of Alzheimer disease: The Nun Study.
controls had higher mean MHT scores than AD JAMA 1997;277:813–817.
controls. The paternal occupation profile was sim- 6. MRC CFAS. Pathological correlates of late-onset demen-
tia in a multicentre, community-based population in En-
ilar in terms of social class (table 1) and consistent
gland and Wales. Lancet 2001;357:169–175.
with the national average, but there may have been 7. Launer LJ, Ross GW, Petrovich H, et al. Midlife blood
subtle differences in actual occupation within the pressure and dementia: the Honolulu-Asia Aging Study.
predominant skilled social class III that explain Neurobiol Aging 2000;21:49–55.
this difference.31 A number of controls may have 8. Kivipelto M, Helkala EL, Hanninen T, Laakso M, Halli-
developed dementia, but this means that the ORs kainen M. Midlife vascular risk factors and Alzheimer’s
disease in later life: longitudinal, population-based study.
more accurately reflect risk relative to the general
BMJ 2001;322:1447–1451.
population rather than a select nondemented
9. Satz P. Brain reserve capacity and symptom onset after
group. Moreover, as indicated in the Introduction, brain injury: a formulation and review of evidence for
clinical diagnostic criteria of AD and VaD cover a threshold theory. Neuropsychology 1993;7:273–295.
spectrum of neuropathology and a proportion of 10. Stern Y. What is cognitive reserve? Theory and research
controls may have had changes within their brains application of the reserve concept. J Int Neuropsychol Soc
without any manifestations of dementia. The dif- 2002;8:448–460.
11. Richards M, Deary IJ. A life course approach to cognitive
ferentiation of VaD from AD using clinical crite-
reserve: a model for cognitive aging and development? Ann
ria is known to be difficult. The diagnostic
Neurol 2005;58:617–622.
accuracy of the ICD-10 criteria is low.32-34 The 12. Whalley LJ, Deary IJ. Longitudinal cohort study of child-
sensitivity of the ICD-10 criteria, as well as others, hood IQ and survival up to age 76. BMJ 2001;322:819–
is very low. Thus while the clinical criteria used 822.
here identified subjects with clinical cerebrovascu- 13. Hart CL, Taylor MD, Davey Smith G, et al. Childhood
lar syndromes, we do not know the extent of the IQ and cardiovascular disease in adulthood: prospective
observational study linking the Scottish Mental Survey
link between the clinical diagnosis of VaD and
1932 and the Midspan studies. Soc Sci Med 2004;59:
cerebrovascular pathology. The matching also
2131–2138.
meant that we had no data to compare effects of 14. Starr JM, Taylor MD, Hart CL, et al. Childhood mental
education and occupation, though childhood IQ ability and blood pressure at midlife: linking the Scottish
is strongly correlated with both in Western societ- Mental Survey 1932 and the Midspan studies. J Hypertens
ies.35 Whether the same pathways linking IQ and 2004;22:893–898.
dementia are relevant to non-Western societies, 15. Whalley LJ, Starr JM, Athawes R, Hunter D, Pattie A,
Deary IJ. Childhood mental ability and dementia. Neurol-
which will see the largest increase in dementia, re-
ogy 2000;55:1455–1459.
quires further research. This is important before
16. Riley KP, Snowden DA, Desrosiers MF, Markesbery WR.
public health measures to reduce dementia inci- Early life linguistic ability, late life cognitive function, and
dence based on preventing vascular pathology are neuropathology: findings from the Nun study. Neurobiol
implemented in less-developed countries. Aging 2005;26:341–347.

Neurology 71 September 30, 2008 1055

 17. Starr JM, Whalley LJ. Differential cognitive outcomes in 27. McDowell I, Xi G, Lindsay J, Tierney M. Mapping the
the Hypertensive Old People in Edinburgh (HOPE) connections between education and dementia. J Clin Exp
study. J Neurol Sci 2005;229–230:103–107. Neuropsychol 2007;29:127–141.
18. SCRE (1933). The Intelligence of Scottish Children: A 28. Stern Y, Alexander GE, Prohovnik I, Mayeux R. Inverse
National Survey of an Age Group. University of London relationship between education and parietotemporal perfu-
Press: London; 1933. sion deficit in Alzheimer’s disease. Ann Neurol 1992;32:
19. Deary IJ, Whiteman MC, Starr JM, Whalley LJ, Fox HC. 371–375.
The impact of childhood intelligence on later life: follow- 29. Cobb JL, Wolf PA, Au R, White R, D’Agostino RB. The
ing up the Scottish Mental Surveys of 1932 and 1947. J effect of education on the incidence of dementia and Alz-
Pers Soc Psychol 2004;1:130–147. heimer’s disease in the Framingham Study. Neurology
20. World Health Organization. The ICD-10 Classification of 1995;45:1707–1712.
Mental and Behavioural Disorders: Clinical Descriptions 30. Fratiglioni L, Grut M, Forsell Y, et al. Prevalence of Alz-
and Diagnostic Guidelines. Geneva: WHO; 1992. heimer’s disease and other dementias in an elderly urban
21. General Register Office for Scotland. Mid-2004 popu- population: relationship with age, sex and education. Neu-
lation estimates. Available at: http://www.gro- rology 1991;41:1886–1892.
scotland.gov.uk/files1/stats/04mype-cahb-booklet- 31. General Register Office of Scotland. Census of Scotland,
revised.pdf 1921, Table 8. HMSO: Edinburgh; 1921.
22. Hofman A, Rocca WA, Brayne C, et al. The prevalence of 32. Gold G, Bouras C, Canuto A, et al. Clinicopathological vali-
dementia in Europe: a collaborative study of 1980 –1990 dation study of four sets of clinical criteria for vascular demen-
findings. Eurodem Prevalence Research Group. Int J Epi- tia. Am J Geriatr Psychiatr 2002;159:82–87.
demiol 1991;20:736–748. 33. Wetterling T, Kanitz RD, Borgis KJ. Comparison of different
23. McGurn B, Starr JM, Topfer JA, et al. Pronunciation of diagnostic criteria for vascular dementia (ADDTC, DSM-IV,
irregular words is preserved in dementia, validating pre- ICD-10, NINDS-AIREN). Stroke 1996;27:30–36.
morbid IQ estimation. Neurology 2004;62:1184–1186. 34. Pohjasvaara T, Mantyla R, Ylikoski R, Kaste M, Erkinjuntti
24. Roman GC, Tatemichi TK, Erkinjuntti T, et al. Vascular T. Comparison of different clinical criteria (DSM-III, AD-
dementia: diagnostic criteria for research studies: report of DTC, ICD-10, NINDS-AIREN, DSM-IV) for the diagnosis
the NINDS AIREN International Workshop. Neurology of vascular dementia National Institute of Neurological Dis-
1993;43:250–260. orders and Stroke-Association Internationale pour la Recher-
25. Stern Y. Cognitive reserve and Alzheimer disease. Alzhei- che et l’Enseignement en Neurosciences. Stroke 2000;31:
mer Dis Assoc Disord 2006;20:112–117. 2952–2957.
26. Deary IJ, Strand SW, Smith P, Fernandes C. Intelligence 35. Deary IJ. Looking Down on Human Intelligence: From
and educational achievement. Intelligence 2007;35:13–21. Psychometrics to the Brain. OUP: Oxford; 2000.

Disagree? Agree? Have a Question? Have an Answer?

Respond to an article in Neurology® through our online Correspondence system:
• Visit www.neurology.org
• Access specific article on which you would like to comment
• Click on “Correspondence: Submit a response” in the content box
• Enter contact information
• Upload your Correspondence
• Press “Send Response”
Correspondence will then be transmitted to the Neurology Editorial Office for review. Correspon-
dence must be received within six weeks of the publication date of the article. Selected correspon-
dence will subsequently appear in the print Journal. See our Information for Authors at
www.neurology.org for format requirements.

1056 Neurology 71 September 30, 2008

 Childhood cognitive ability and risk of late-onset Alzheimer and vascular dementia
Brian McGurn, Ian J. Deary and John M. Starr
Neurology 2008;71;1051-1056 Published Online before print June 25, 2008
DOI 10.1212/01.wnl.0000319692.20283.10

This information is current as of June 25, 2008

Updated Information & including high resolution figures, can be found at:
Services http://www.neurology.org/content/71/14/1051.full.html

References This article cites 29 articles, 11 of which you can access for free at:
http://www.neurology.org/content/71/14/1051.full.html##ref-list-1
Citations This article has been cited by 4 HighWire-hosted articles:
http://www.neurology.org/content/71/14/1051.full.html##otherarticles
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
Alzheimer's disease
http://www.neurology.org//cgi/collection/alzheimers_disease
Case control studies
http://www.neurology.org//cgi/collection/case_control_studies
Risk factors in epidemiology
http://www.neurology.org//cgi/collection/risk_factors_in_epidemiology

Vascular dementia
http://www.neurology.org//cgi/collection/vascular_dementia
Permissions & Licensing Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/misc/about.xhtml#permissions
Reprints Information about ordering reprints can be found online:
http://www.neurology.org/misc/addir.xhtml#reprintsus

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright . All rights reserved. Print ISSN: 0028-3878.
Online ISSN: 1526-632X.