- Email: [email protected]
Cognitive ability in young adulthood and risk of dementia in a cohort of Danish men, brothers, and twins
Alzheimer’s & Dementia - (2017) 1-9
Featured Article
Cognitive ability in young adulthood and risk of dementia in a cohort of Danish men, brothers, and twins Merete Oslera,b,c,*, Gunhild T. Christensena,b,c, Ellen Gardeb,d,e, Erik L. Mortensenb,c,e, Kaare Christensenc,f,g a
Research Center for Prevention and Health, Rigshospitalet-Glostrup, University of Copenhagen, Glostrup, Denmark b Department of Public Health, University of Copenhagen, Denmark c Department of Public Health, Danish Aging Research Center, University of Southern Denmark, Odense, Denmark d Danish Research Centre for Magnetic Resonance, Centre for Functional and Diagnostic Imaging and Research, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark e Center for Healthy Aging, University of Copenhagen, Copenhagen, Denmark f Department of Clinical Genetics, Odense University Hospital, Odense, Denmark g Department of Clinical Biochemistry and Pharmacology, Odense University Hospital, Odense, Denmark
Abstract
Introduction: We examined the association between cognitive ability in young adulthood and dementia in Danish men, brothers, and male twins. Methods: In total, 666,986 men born between 1939 and 1959 were identified for dementia diagnosis in national registries from 1969 to 2016. The association between cognitive ability from draft board examination and dementia was examined using Cox regression. Results: During a 44-year follow-up, 6416 (0.96%) men developed dementia, 1760 (0.26%) and 970 (0.15%) of which were classified as Alzheimer’s and vascular dementia, respectively. Low cognitive ability was associated with increased risk of dementia (hazard ratio [HR]per SD decrease 1.33 [95% confidence interval {CI} 5 1.30–1.35]) with the strongest associations for vascular dementia (HRper SD decrease 1.47 [95% CI 5 1.31–1.56]) and a weaker for Alzheimer’s disease (HRper SD decrease 1.07 [95% CI 5 1.03–1.13]). The intrabrother and twin analyses (taking shared family factors into account) showed attenuated risk estimates but with wide CIs. Discussion: Low early-life cognitive ability increases the risk of dementia before the age of 78 years. The association is partly explained by shared family factors. Ó 2017 the Alzheimer’s Association. Published by Elsevier Inc. All rights reserved.
Keywords:
Cohort study; Twin study; Cognitive ability; Dementia; Alzheimer’s disease; Vascular dementia
1. Introduction Dementia is a leading cause of disability, and with an increasing elderly population worldwide, the prevalence of dementia is expected to increase, despite reductions in incidence [1–4]. Alzheimer’s disease (AD) is the most common cause of dementia in the elderly, followed by vascular
The authors have declared that no conflict of interest exists. *Corresponding author. Tel.: 14538633780; Fax: 14538633977. E-mail address: [email protected]
dementia (VaD). The increasing awareness of shared modifiable risk factors such as cardiovascular risk factors [3] calls for an identification of such risk factors and effective strategies for prevention of dementia. In a life-course perspective, general early-life cognitive ability has been proposed as a potential risk factor for dementia. Low early-life cognitive ability reflects prenat al brain development [5–7] and is associated with subsequent morbidity and mortality [8–11]. This might increase the risk of dementia, which acts not only through cognitive reserve but also through poorer lifestyle and comorbidity, potentially influencing thrombotic and atherosclerotic processes and might
http://dx.doi.org/10.1016/j.jalz.2017.04.003 1552-5260/Ó 2017 the Alzheimer’s Association. Published by Elsevier Inc. All rights reserved. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
2
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
increase the risk of vascular lesions [6]. However, the influence of early-life cognitive ability and lifestyle on risk of developing AD and VaD is scarcely investigated, mainly due to the lack of access to information on early-life cognitive ability. In the study of 93 nuns from Milwaukee, USA, linguistic ability at the age of 22 years was associated with cognitive test scores and a diagnosis of AD at the age of 75 to 95 years [12]. In the Scottish 1921 birth cohort, 50 patients with dementia diagnosed after the age of 65 years had lower mental ability at the age of 11 years compared with controls, but this was not the case for 59 patients with early-onset dementia [13]. Furthermore, analysis of the Scottish birth cohort also showed that the 31 cases with VaD had lower early-life mental ability than matched controls, but there was no significant difference for 63 cases with AD [14]. More recently, two studies based on the Swedish Military Conscription Register [15,16] have shown that low cognitive ability in young adulthood was associated with dementia before the age of 60 years (dementia cases: n 5 487 [15] and n 5 657 [16]). Family studies have suggested that more than 50% of the between-subject variation in both early-life cognitive ability [17] and AD can be ascribed to genetic factors [17–19]. Environmental factors shared within families and attributed to cognitive ability include parental lifestyle and socioeconomic conditions [20]. Twin and sibling studies offer a unique design for controlling, in part, for shared genetic and familial factors [21,22] by taking advantage of the fact that twins and full siblings are almost always brought up together and that monozygotic (MZ) twins are perfectly matched genetically, whereas dizygotic (DZ) twins and full siblings share on average 50% of their segregating genes. To date, no studies have investigated the impact of shared familial factors on the association between earlylife cognitive ability and late-life dementia. The objectives of the study were to examine the association between cognitive ability in young adulthood and AD or VaD diagnosis and age of onset in men. We also wanted to assess whether this association could be replicated within brothers and male twin pairs. We hypothesize that low early-life cognitive ability increases the risk of dementia, and that familial factors including genetic constitution account for a considerable part of the expected association.
2. Methods 2.1. Study population The Danish Conscription Database (DCD) comprises 728,158 Danish men born between 1939 and 1959, who were conscripted for mandatory military service from 1957 until 1984 and were alive on April 1, 1968 when the Danish Civil Registration System was established. Through data linkage with the Danish Twin Registry and the Danish Civil Registration System, 14,408 male twins (4108 intact male twin pairs n 5 8216) and 81,002 brothers were
identified in the DCD. Of the 728,158 men, 63,172 (0.9 %) were excluded; 52,843 had no conscript examination data, whereas 5691, 1181, and 1458 had no information on cognitive ability, education, or height. Available for analysis were 666,986 men including 74,761 brothers and 7310 male twins (n 5 3655 pairs with information on zygosity for 1085 MZ and 2226 DZ pairs) with complete information on all data. The missing observations were mainly due to the lack of registry information for men who volunteered for military service or had been exempted from the draft board examination because of the medical conditions such as intellectual disability, psychiatric disorders, epilepsy, or type 1 diabetes. The percentage of missing observation was highest (8.2%) for men born after 1955 and lowest (5.9%) for those born before 1944. The DCD has been described in detail, previously [23]. All data linkages and analyses were approved by the Danish Data Protection Agency. 2.2. Assessment of cognitive ability Cognitive ability was assessed by the draft board intelligence test, the Børge Prien Prøve (BPP), which has been described previously [23,24] and used in other studies [9]. It is a group-administered paper and pencil test comprising four subtests (letter matrices, 19 items; verbal analogies, 24 items; number series, 17 items; and geometric figures, 18 items) to be completed within 45 minutes. This version of BPP has been used without changes since 1957. The numbers of correct answers in the four subtests are summed to a total score with a 0 to 78 range. The total BPP score has been shown to correlate with the full-scale Wechsler Adult Intelligence Scale (r 5 0.82) [25]. 2.3. Dementia outcome measures Information on any admission to a psychiatric or somatic ward from 1969 or 1977, respectively, until 2016 was obtained by linking the DCD to the Danish Psychiatric Central Registry and the Danish National Patient Registry using the person identification number as a key. These registers hold individual-level data on type of patient contact (inpatients, and, from 1995 onward, emergency and outpatients), diagnosis, and date of admission for all hospital admissions [26]. Mortality from all and specific causes of death was followed from April 1968 until April 2016 in the Danish Cause of Death Register. Hospital diagnosis and registered causes of death have been classified according to 8th revision of the International Classification of Disease (ICD-8) for the period 1969 to 1993, and according to the ICD-10 from 1994 onward. In the present study, the outcome of interest was the first hospital discharge or death from a main diagnosis of dementia (ICD-8: 290.00–290.99 and ICD-10: F00.0–F03.9; G30.0–G30.9) from 1969 until 2016. Dementia cases were subdivided into AD (ICD-8: 290.10 and ICD-10: F00.0–F00.9; G30.0–G30.9) and VaD (ICD-10: F01.0–F01.9) cases.
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
2.4. Validity of dementia diagnosis The validity of dementia diagnosis obtained from the previously mentioned registers has been assessed in two independent studies of patients admitted in 2003 [27] and 2008 [28]. They revealed that 70% [28] and 83% [27] of dementia cases diagnosed by an external rater according to ICD-10 confirmed the diagnosis recorded in the register [27,28] with the lowest validity in younger populations and for VaD. The validity of diagnosis based on ICD-8 has not been examined. In the DCD cohort, 663 and 5563 men were registered with a dementia diagnosis defined by either ICD-8 or ICD-10. 2.5. Covariables From the DCD, height, education, and any registered admission to a psychiatric hospital at study entry were included as covariables because these variables have been shown to be associated with early-life cognitive ability and risk of dementia [8,29]. Educational level reported by the men at conscription was coded into three categories comprising low (7th–9th grade), medium (vocational training or 10th–11th grade), and high (12th grade or more advanced) educational levels. Height was measured without shoes at the physical examination by a physician. 2.6. Statistical analyses
board examination (for men born 1951–1959 conscripted in 1969–1984; age 19 years). Follow-up was terminated at the age of first admission, death, emigration, or end of followup (April 1, 2016; age 57–78 years), whichever came first (Fig. 1). A few subjects (n 5 4) diagnosed with dementia before the start of follow-up were excluded. To examine the shape of the association between cognitive ability and dementia, we divided the BPP scores into six groups of equal size. This indicated a linear trend, but there were few outcomes in some strata. To obtain categories with sufficient number of cases and for comparability with previous studies [15], the BPP was grouped in tertiles. The BPP was also analyzed as a continuous variable. In multiple regression models, height, education, and any psychiatric hospital admission at study entry were included as covariables, both one at a time and together. Any interaction of cognitive ability with height, education, or birth cohort was examined comparing models with and without an interaction term using likelihood ratio tests, but there was no indication of significant multiplicative interaction. For siblings, we first examined how often the brother/co-twin with the lowest cognitive score was the first to be diagnosed with dementia compared with his higher scoring brother/co-twin (only including the first two brothers from families with more than two brothers). Secondly, we conducted a standard cohort analysis treating each twin or brother as individuals, taking the interdependence of observations within siblings into account by including a cluster term. In the intrasibling and twin analyses, a variable identifying membership of a twin pair or sibling group was included as a stratum variable, fixing the baseline hazard within siblings, but allowing it to differ between individuals who were not siblings. Thus, the hazard ratios (HRs) were estimated conditional on the family-level intercept. This corresponds to conditional logistic regression, and made it possible to
web 4C=FPO
The influence of cognitive ability in young adulthood on the incidence of dementia was analyzed using Cox regression with age as the underlying time scale. Person-years (py) of follow-up for a dementia diagnosis was accumulated from age on January 1, 1969 (for men born 1939–1950 conscripted in 1957–1968; age 19–30 years) or on the date of the draft
3
Fig. 1. Venn diagram with age and time for follow-up of the men born between 1939 and 1959 in the Danish Conscription Database. Abbreviations: ICD-8, 8th revision of the International Classification of Disease; ICD-10, 10th revision of the International Classification of Disease. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
4
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
compare siblings while keeping common factors fixed (i.e., family factors shared within siblings). We took subjects who died before they could have been admitted (competing risk) into account using Fine-Gray competing risk regression by the “stcrreg” command in STATA. This provides “subdistribution HRs,” which can be used as an alternative to Cox regression to determine whether the observed associations can be attributed to competing risks [30]. The proportional hazard assumption was examined by inspection of cumulative hazard plots. This showed that around the age of 60 years, the difference between curves for cognitive ability diminished, and, consequently, analyses for early- and late-onset dementia were split at that age. Subsequently, the test of proportional hazards revealed no violation of the assumptions. Finally, we examined the association between cognitive ability and dementia defined based on ICD-8 or ICD-10 in men born between 1939 and 1949 followed from the age of 45 to 65 years and in men born between 1950 and 1959 followed through the same ages. All analyses were carried out in STATA, version 12. 3. Results 3.1. Incidence of dementia diagnosed during follow-up The 666,986 men were followed for a median of 44.1 years (27.8 million py), and during that time, 6416 (0.96%) developed dementia (incidence rate 5 23.4 [95% confidence interval {CI} 5 22.8–23.9] per 100,000 py). Of these, 1760 (0.26%) and 970 (0.15%) were classified as AD or VaD, respectively, whereas 3686 dementia cases had an unspecified diagnosis (ICD-10: F03.9). 3.2. Cognitive ability and risk of dementia
3.3. Cognitive ability and risk of dementia in brothers and twins Among the 74,759 brothers and 7310 male twins, 340 (0.45%) and 60 (0.82%) developed dementia during follow-up, respectively. Supplementary Table 1 presents the intrasibling discordance in young adult cognitive ability for brothers and twins and shows that percentage of pairs with more than 1 standard deviation (SD) (.11 points difference) discordance on cognition was higher in brothers than in twins. In pairs (n 5 292) of brothers where one or both had developed dementia, the proportion of pairs in which the brother with the lowest score was diagnosed with
web 4C=FPO
Figs. 2–4 and Table 1 show the distribution of dementia diagnoses in relation to cognitive ability in young adulthood
and the corresponding crude and adjusted HRs. Fig. 2 shows that low cognitive ability in young adulthood was associated with an increased risk of dementia (HRlowest versus highest tertile 1.74 [95% CI 5 1.60–1.88] or HRper SD decrease 1.33 [95% CI 5 1.30–1.35]) with the strongest associations observed for VaD (Figs. 3 and 4). A total of 2538 (39.6%) of dementia cases were diagnosed before the age of 60 years, of these, 281 (15.0%) as AD and 298 (30.7%) as VaD, and cognitive ability seemed to be more strongly associated with the early-onset dementia (Fig. 2). Table 1 also shows that dementia was more frequent in men with low education or height, and both factors were associated with increased risk of dementia (HRlow versus high education 1.50 [95% CI 5 1.38–1.63] or HRlow versus high height 1.62 [95% CI 5 1.52–1.73]). However, adjustment for education and/ or height at conscription only slightly influenced the estimates (Figs. 2–4). Few men (0.3%) had been admitted to a psychiatric hospital at study entry, and no association with dementia risk was detected (HR 1.42 [95% CI 5 0.82– 2.44]) or influenced the association between cognitive ability and dementia. Taking competing mortality risk into account did not change the conclusions.
Fig. 2. Number with dementia and associations (HRs and 95% CIs) in relation to cognitive ability in young adulthood among 666,934 Danish men born between 1939 and 1959. Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
5
web 4C=FPO
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
Fig. 3. Number with Alzheimer’s disease and associations (HRs and 95% CIs) in relation to cognitive ability in young adulthood among 666,934 Danish men born between 1939 and 1959. Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation.
(HRper SD decrease 1.36[95% CI 5 1.08–1.73]) (Table 2), which was close to the estimates for the overall sample and for men with no brothers and those with no parental identification. The adjusted intrasibling analyses, which took into account shared family factors, indicated slightly attenuated risk estimates for brothers (HRper SD decrease 1.32 [95% CI 5 1.04–1.64]). In male twin pairs, the HRs were more attenuated but also imprecisely estimated
web 4C=FPO
dementia first was 57%. This percentage was higher than the 50% (assuming no difference) but was not related to the difference in cognitive ability within brother pairs (being 60%, 50%, and 59% in pairs with less than ½ SD, ½–1 SD, and more than 1 SD difference in cognitive scores, respectively). In the adjusted standard cohort analysis, we found a relation between cognitive ability and dementia in both brothers (HRper SD decrease 1.47 [95% CI 5 1.31–1.62]) and twins
Fig. 4. Number with vascular dementia and associations (HRs and 95% CIs) in relation to cognitive ability in young adulthood among 666,934 Danish men born between 1939 and 1959. Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
6
Table 1 The number (n) with dementia in relation to cognitive ability, education, and height in early adulthood among 666,934 Danish men born between 1939 and 1959 Total cohort
Total Cognitive ability Lowest Middle Highest Education Lowest Middle Highest Height Lowest Middle Highest Birth year 1939–1944 1945–1949 1950–1954 1955–1959 Prior psychiatric admission No Yes
Men with parental identification
Male twins
N (%)
Dementia n (%)
Alzheimer’s disease n (%)
Vascular dementia n (%)
N (%)
Dementia n (%)
N (%)
Dementia n (%)
666,986 (100)
6416 (0.96)
1760 (0.26)
970 (0.15)
245,998 (100)
1235 (0.50)
12,941
140 (1.08)
237,959 (35.68) 219,243 (32.87) 209,784 (31.45)
3142 (1.32) 1853 (0.85) 1421 (0.68)
742 (0.31) 518 (0.24) 500 (0.24)
404 (0.21) 297 (0.14) 169 (0.08)
71,859 (24.21) 84,961 (34.54) 89,178 (36.25)
587 (0.802) 360 (0.42) 288 (0.32)
5808 (44.86) 4009 (31.02) 3124 (24.12)
84 (1.45) 34 (0.85) 22 (0.70)
171,699 (25.74) 381,495 (57.20) 113,792 (17.06)
2042 (1.19) 3624 (0.95) 750 (0.66)
475 (0.28) 1015 (0.27) 270 (0.24)
346 (0.20) 525 (0.14) 99 (0.09)
54,001 (22.36) 135,709 (55.17) 55,288 (22.48)
404 (0.73) 644 (0.47) 187 (0.34)
3804 (29.41) 7350 (56.78) 1787 (13.81)
55 (1.45) 75 (1.02) 10 (0.56)
242,277 (36.32) 235,729 (35.34) 188,980 (28.33)
3012 (1.24) 2128 (0.90) 1276 (0.68)
781 (0.32) 585 (0.25) 389 (0.21)
495 (0.20) 325 (0.14) 150 (0.08)
72,720 (29.56) 87,852 (35.71) 85,426 (34.73)
508 (0.70) 405 (0.46) 322 (0.38)
5425 (41.92) 4442 (34.34) 3074 (23.84)
77 (1.42) 41 (0.92) 22 (0.72)
177,647 (26.63) 186,155 (27.91) 162,156 (24.31) 141,028 (21.15)
3385 (1.91) 1829 (0.98) 803 (0.50) 399 (0.28)
1112 (0.63) 441 (0.24) 143 (0.09) 64 (0.05)
540 (0.30) 270 (0.15) 101 (0.06) 52 (0.04)
15,601 (6.34) 23,069 (9.38) 75,303 (30.61) 132,025 (53.67)
286 (1.83) 240 (1.04) 352 (0.47) 357 (0.50)
3353 (25.91) 3810 (29.44) 3009 (23.31) 2769 (21.34)
81 (2.42) 34 (0.89) 18 (0.60) 7 (0.25)
666,048 (99.71) 1938 (0.29)
6403 (0.96) 13 (0.69)
1760 (0.26) 0 (0.00)
968 (0.14) 2 (0.10)
244,690 (99.47) 1308 (0.53)
1230 (0.50) 5 (0.38)
12,209 (99.70) 39 (0.30)
138 (1.07) 2 (5.13)
(HRper SD decrease 1.09 [95% CI 5 0.55–2.13]). When the twin analysis was made separately for MZ and DZ twin pairs, the estimates for MZ (HRper SD decrease 0.30 [95% CI 5 0.04–2.83]) and DZ (HRper SD decrease 1.09 [95% CI 5 0.50–2.32]) twin pairs became very imprecise as they were based on only 14 and 51 cases with dementia.
with overlapping CIs (HRICD-8: lowest versus highest tertile 3.64 [95% CI 5 2.78–4.76] and HRICD-10: lowest versus highest tertile 2.81 [95% CI 5 2.16–3.66]). Finally, we examined dementia outcomes in men with complete information as compared to those with partly and complete missing data. This showed that the latter group had higher risk of dementia (HR 1.70 [95% CI 5 1.51–1.91]), whereas there was no difference for those with missing data on single variables.
3.4. Supplementary analyses We also analyzed data including dementia coded as secondary diagnosis in the outcome definition. This showed the same patterns of associations as reported previously. When the analyses were repeated on diagnosis based on ICD-8 or ICD-10 in a subsample of men followed during the same age span, the estimates seemed to be somewhat stronger for diagnoses based on ICD-8 than ICD-10, but
4. Discussion 4.1. Main findings In this cohort of 666,986 men followed for a median of 44 years from about the age of 19 years, 1 SD decrease in cognitive ability measured at baseline was associated with a 33% higher risk of dementia. The risk estimates were
Table 2 Associations (adjusted hazard ratio and 95% CIs) between cognitive ability in young adulthood and subsequent risk of dementia among 666,934 Danish men born 1939–1959, showing results from standard cohort and intrasibling and twin analyses (168 and 34 affected pairs) Standard cohort analysis
Dementia cases/total number Cognitive ability Lowest Middle Highest Per SD increase
Intrasibling and twin analysis No parental information
No brother
Brothers
Twins
5181/420,988
895/171,237
340/74,761
69/7310
1.63 (1.50–1.77) 1.16 (1.07–1.26) 1 1.33 (1.28–1.37)
2.05 (1.67–2.53) 1.26 (1.06–1.54) 1 1.47 (1.30–1.61)
2.34 (1.65–3.31) 1.26 (0.97–1.78) 1 1.47 (1.31–1.62)
1.25 (0.53–2.91) 1.20 (0.52–2.80) 1 1.36 (1.08–1.73)
Abbreviations: CI, confidence interval; SD, standard deviation. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
Brothers
Twins
1.89 (1.05–3.38) 1.11 (0.65–1.89) 1 1.32 (1.04–1.64)
1.71 (0.34–8.48) 2.69 (0.64–11.21) 1 1.09 (0.55–2.13)
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
stronger for VaD than AD and for early- versus late-onset dementia of all subtypes. The associations were slightly attenuated when adjusted for education and/or height measured at the age of 19 years. In the intrasibling and twin analysis, the associations between cognitive ability and dementia were attenuated indicating that family factors explain part of the association, although the estimates for brothers were imprecisely estimated, especially for the twins. 4.2. Comparison with other studies Our results are in accordance with previous studies on dementia with early [16] and late onset [14,16], but in contrast, the Scottish birth cohort showed no association between cognitive ability at the age of 11 years and early-onset dementia [13]. However, our study is comparable to the study of Swedish conscripts [16] reporting both an association between cognitive ability in early adulthood and early-onset dementia. The relations between cognitive ability in early life and AD and VaD were also examined in the Scottish birth cohort, and in accordance with our findings, the study showed an association between low cognitive ability at the age of 11 years and VaD. However, there was no relation with AD, possibly because the number of AD patients (n 5 63) was insufficient to detect a small difference in cognitive ability between patients and controls. In the present study, education in young adulthood had only minor influence on the association between cognitive ability and dementia outcomes. Similarly, in the Nun Study [12], linguistic ability at the age of 22 years was associated with cognitive ability 58 years later even when adjusted for education. In the Swedish conscript cohort, adjustment for education at the age of 18 years had the largest, although still modest, attenuating influence on the association between cognitive ability and dementia, whereas parental education and cardiovascular fitness did not influence the risk estimates [16]. The Swedish researchers also explored the influence of neurological and psychiatric disorders emerging up to 15 years after conscription on the risk estimates. Although the exclusion of subjects with these disorders did not seem to influence the association between cognitive ability and dementia, it is not clear how the time dependency of these comorbidities was accounted for in the analyses. In our study, adjustment for prior psychiatric hospital admission had no influence on the association between cognitive ability and dementia. Our study is the first to suggest that family environment and genetic constitution might explain part of the association, but risk estimates were not very precise as they were based on the subsample of 74,761 brothers and 7310 male twins with 340 and 69 dementia cases, respectively. The information on parental identity was obtained from the Danish Civil Registration System and is almost complete for persons born after 1955, but the availability decreases rapidly for those born earlier from about 10% for those born in 1950 to less than 1% for those born before 1940
7
[31]. Consequently, we had missing information on parents for more than two-thirds of cohort members, but sensitivity analyses showed no difference in the pattern of associations between individuals with and without parental information. Some (3.3%) of the brothers represented a twin pair, but since the two populations of brothers and twins were sampled from the Danish Civil Registration System and the Danish Twin Registry, with different degrees of completeness, we decided to analyze two independent substudies. 4.3. Study strength and limitations The strength of the present analysis is that it was based on a relatively large sample compared with the Scottish studies [13,14] and that the sample covered older birth cohorts (1939–1959) than the Swedish conscription cohort (1950–1987) [16]. Furthermore, cognitive ability was assessed in young adulthood where it can be expected to be relatively stable. In our study, outcome information was retrieved from national registers with sensitivity measures .70% [26–28]. The diagnosis of dementia in young persons might be less valid because of a different prevalence of subtypes and symptom profile and a range of mainly psychiatric or neurological conditions that may be misdiagnosed as dementia. Thus, we cannot exclude that the observed, stronger association between cognitive ability and dementia diagnosed before the age of 60 years could reflect that low cognitive ability was misinterpreted as dementia in younger persons. On the other hand, information on cognitive ability was not available for those exempted from conscription because of mental disability and severe psychiatric disorder, and our sensitivity analysis showed that subjects with missing data had higher risk of dementia than those available for the present study. This suggests that our estimates of the relation between cognitive ability and dementia might not be explained by confounding from such disorders. Our study covered 20 birth cohorts followed for outcomes for a median of 44 years from ages 18 to 30 years. Thus, follow-up ended when cohort members were between 56 and 76 years and, consequently, our study does not include old-age dementia outcomes. During this time period, ICD coding has changed from ICD-8 to ICD-10, and the coverage of the hospital registers has increased because of the inclusion of outpatient contacts from 1995. However, we did not find any major difference in the risk estimates for the association between cognitive ability and earlyonset dementia across birth cohorts or in substudies with dementia defined based on ICD-8 and ICD-10, respectively. Dementia is more prevalent in women, and agedependent differences in the prevalence between men and women have been reported [32]. Our cohort consisted of men only, and the results are not necessarily generalizable to women. However, results from the Nun Study and the Scottish birth cohorts support a relationship between
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
8
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
cognitive ability in the young and risk of dementia also in women. Although the study is large, the analyses, especially those on siblings and twins, are still affected by lack of statistical power.
4.4. Possible mechanism and implications of our findings Mechanisms linking cognitive ability in early life with later health outcomes [5,14,19] include confounding from genetic constitution, familial factors, and poor health during early childhood or influences mediated through poor lifestyle and health. In our study, education, height, and early psychiatric comorbidity, which are closely related to deprivation in early life, were not important confounders, suggesting that attenuation of the association in the intrasibling and twin analyses should rather be attributed to other family factors, such as genetic and environmental factors contributing to cognitive reserve as a compensatory mechanism for coping with brain pathology [33]. The relatively strong association between early-life cognitive ability and VaD suggests a combined risk of limited cognitive reserve and poor lifestyle and health. Studies report on an association of early-life cognitive abilities with risk of cardiovascular events [34] and cardiovascular risk factors are not only associated with an increased risk of VaD but also AD [35]. The Flynn effect, that is, an observation of secular increases in young adult cognitive scores across a range of populations, is strong evidence for an important environmental effect, including lifestyle. This effect has occurred far too quickly to be explained purely by changes in the gene pool or other biological effects [23,36], but it may have contributed to the recent decline in the incidence of dementia observed in Western populations [2,3]. In conclusion, low cognitive ability in young adulthood seems to be associated with higher risk of all types of dementia. Consequently, early-life cognitive ability may provide a further modifiable risk factor for dementia. Thus, we suggest that prevention of dementia should be initiated early in life and attempts should be made to preserve cognitive ability especially among those with low ability.
Acknowledgments The work was supported by the Danish Medical Research Council [grant number 09-063599 and 09-069151] and The Velux Foundations [grant number 31205].
Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jalz.2017.04.003.
RESEARCH IN CONTEXT
1. Systematic review: In a study of 93 American nuns, linguistic ability at the age of 22 years was associated with late-life Alzheimer’s disease (AD). Later, the Scottish birth cohort showed that low mental ability at the age of 11 years increased the risk of vascular dementia (VaD) but not of AD. Recently, Swedish conscription data have shown that low cognitive ability was associated with early-onset dementia. We are the first to examine whether an association between early-life cognitive ability and dementia is replicated within brothers and male twins. 2. Interpretation: In accordance with the previous studies, low cognitive ability was associated with increased risk of dementia before the age of 75 years. The novel findings were the stronger association for VaD than for AD, and the contribution of environmental factors shared within families to the association between low early-life cognitive ability and dementia. 3. Future direction: Future research should further explore the factors mediating the association between cognitive ability and dementia and any differential influence of familial factors on the risk of VaD and AD.
References [1] Wu YT, Fratiglioni L, Matthews FE, Lobo A, Breteler MM, Skoog I, et al. Dementia in Western Europe: epidemiological evidence and implications for policy making. Lancet Neurol 2016;15:116–24. [2] Matthews FE, Stephan BC, Robinson L, Jagger C, Barnes LE, Arthur A, et al. A two decade dementia incidence comparison from the Cognitive Function and Ageing Studies I and II. Nat Commun 2016;7:11398. [3] Satizabal CL, Beiser AS, Chouraki V, Chene G, Dufouil C, Seshadri S. Incidence of Dementia over three decades in the Framingham Heart Study. N Engl J Med 2016;374:523–32. [4] Wimo A, Guerchet M, Ali GC, Wu YT, Prina AM, Winblad B, et al. The worldwide costs of dementia 2015 and comparisons with 2010. Alzheimers Dement 2017;13:1–7. [5] Papachristou E, Frangou S, Reichenberg A. Expanding conceptual frameworks: life course risk modelling for mental disorders. Psychiatry Res 2013;206:140–5. [6] Richards M, Deary IJ. A life course approach to cognitive reserve: a model for cognitive aging and development? Ann Neurol 2005; 58:617–22. [7] Lynch J, Smith GD. A life course approach to chronic disease epidemiology. Annu Rev Public Health 2005;26:1–35. [8] Osler M, Lawlor DA, Nordentoft M. Cognitive function in childhood and early adulthood and hospital admission for schizophrenia and
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
[9]
[10]
[11] [12]
[13] [14]
[15]
[16]
[17]
[18] [19]
[20]
[21]
[22]
bipolar disorders in Danish men born in 1953. Schizophr Res 2007; 92:132–41. Christensen GT, Mortensen EL, Christensen K, Osler M. Intelligence in young adulthood and cause-specific mortality in the Danish Conscription Database – a cohort study of 728,160 men. Intelligence 2016;59:64–71. Calvin CM, Deary IJ, Fenton C, Roberts BA, Der G, Leckenby N, et al. Intelligence in youth and all-cause-mortality: systematic review with meta-analysis. Int J Epidemiol 2011;40:626–44. Wraw C, Deary IJ, Gale CR, Der G. Intelligence in youth and health at age 50. Intelligence 2015;53:23–32. Snowdon DA, Kemper SJ, Mortimer JA, Greiner LH, Wekstein DR, Markesbery WR. Linguistic ability in early life and cognitive function and Alzheimer’s disease in late life. Findings from the Nun Study. JAMA 1996;275:528–32. Whalley LJ, Starr JM, Athawes R, Hunter D, Pattie A, Deary IJ. Childhood mental ability and dementia. Neurology 2000;55:1455–9. McGurn B, Deary IJ, Starr JM. Childhood cognitive ability and risk of late-onset Alzheimer and vascular dementia. Neurology 2008; 71:1051–6. Nordstrom P, Nordstrom A, Eriksson M, Wahlund LO, Gustafson Y. Risk factors in late adolescence for young-onset dementia in men: a nationwide cohort study. JAMA Intern Med 2013;173:1612–8. Nyberg J, Aberg MA, Schioler L, Nilsson M, Wallin A, Toren K, et al. Cardiovascular and cognitive fitness at age 18 and risk of early-onset dementia. Brain 2014;137:1514–23. Polderman TJ, Benyamin B, de Leeuw CA, Sullivan PF, van Bochoven A, Visscher PM, et al. Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nat Genet 2015; 47:702–9. Deary IJ, Johnson W, Houlihan LM. Genetic foundations of human intelligence. Hum Genet 2009;126:215–32. Gatz M, Reynolds CA, Fratiglioni L, Johansson B, Mortimer JA, Berg S, et al. Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry 2006;63:168–74. Nisbett RE, Aronson J, Blair C, Dickens W, Flynn J, Halpern DF, et al. Intelligence: new findings and theoretical developments. Am Psychol 2012;67:130–59. McGue M, Osler M, Christensen K. Causal inference and observational research: the utility of twins. Perspect Psychol Sci 2010; 5:546–56. D’Onofrio BM, Class QA, Rickert ME, Sujan AC, Larsson H, KujaHalkola R, et al. Translational epidemiologic approaches to understanding the consequences of early-life exposures. Behav Genet 2016;46:315–28.
9
[23] Christensen GT, Molbo D, Angquist LH, Mortensen EL, Christensen K, Sorensen TI, et al. Cohort profile: The Danish Conscription Database(DCD): a cohort of 728,160 men born from 1939 through 1959. Int J Epidemiol 2015;44:432–40. [24] Teasdale TW. The Danish draft board’s intelligence test, Borge Priens Prove: psychometric properties and research applications through 50 years. Scand J Psychol 2009;50:633–8. [25] Mortensen EL, Reinisch JM, Teasdale TW. Intelligence as measured by the WAIS and a military draft board group test. Scand J Psychol 1989;30:315–8. [26] Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sorensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7:449–90. [27] Phung TK, Andersen BB, Hogh P, Kessing LV, Mortensen PB, Waldemar G. Validity of dementia diagnoses in the Danish hospital registers. Dement Geriatr Cogn Disord 2007;24:220–8. [28] Salem LC, Andersen BB, Nielsen TR, Stokholm J, Jorgensen MB, Rasmussen MH, et al. Overdiagnosis of dementia in young patients a nationwide register-based study. Dement Geriatr Cogn Disord 2012;34:292–9. [29] Vagero D, Modin B. Commentary: the associations between height, cognition, and education and their relevance for health studies. Int J Epidemiol 2006;35:663–4. [30] Andersen PK, Geskus RB, de Witte T, Putter H. Competing risks in epidemiology: possibilities and pitfalls. Int J Epidemiol 2012; 41:861–70. [31] Pedersen CB, Gotzsche H, Moller JO, Mortensen PB. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 2006;53:441–9. [32] Lobo A, Launer LJ, Fratiglioni L, Andersen K, Di Carlo A, Breteler MM, et al. Prevalence of dementia and major subtypes in Europe: a collaborative study of population-based cohorts. Neurologic Diseases in the elderly research group. Neurology 2000;54:S4–9. [33] Stern Y. Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 2012;11:1006–12. [34] Dobson KG, Chow CH, Morrison KM, Van Lieshout RJ. Associations between childhood cognition and cardiovascular events in adulthood: a systematic review and meta-analysis. Can J Cardiol 2017;33:232–42. [35] Barnes DE, Yaffe K. The projected effect of risk factor reduction on Alzheimer’s disease prevalence. Lancet Neurol 2011;10:819–28. [36] Trahan LH, Stuebing KK, Fletcher JM, Hiscock M. The Flynn effect: a meta-analysis. Psychol Bull 2014;140:1332–60.
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
Featured Article
Cognitive ability in young adulthood and risk of dementia in a cohort of Danish men, brothers, and twins Merete Oslera,b,c,*, Gunhild T. Christensena,b,c, Ellen Gardeb,d,e, Erik L. Mortensenb,c,e, Kaare Christensenc,f,g a
Research Center for Prevention and Health, Rigshospitalet-Glostrup, University of Copenhagen, Glostrup, Denmark b Department of Public Health, University of Copenhagen, Denmark c Department of Public Health, Danish Aging Research Center, University of Southern Denmark, Odense, Denmark d Danish Research Centre for Magnetic Resonance, Centre for Functional and Diagnostic Imaging and Research, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark e Center for Healthy Aging, University of Copenhagen, Copenhagen, Denmark f Department of Clinical Genetics, Odense University Hospital, Odense, Denmark g Department of Clinical Biochemistry and Pharmacology, Odense University Hospital, Odense, Denmark
Abstract
Introduction: We examined the association between cognitive ability in young adulthood and dementia in Danish men, brothers, and male twins. Methods: In total, 666,986 men born between 1939 and 1959 were identified for dementia diagnosis in national registries from 1969 to 2016. The association between cognitive ability from draft board examination and dementia was examined using Cox regression. Results: During a 44-year follow-up, 6416 (0.96%) men developed dementia, 1760 (0.26%) and 970 (0.15%) of which were classified as Alzheimer’s and vascular dementia, respectively. Low cognitive ability was associated with increased risk of dementia (hazard ratio [HR]per SD decrease 1.33 [95% confidence interval {CI} 5 1.30–1.35]) with the strongest associations for vascular dementia (HRper SD decrease 1.47 [95% CI 5 1.31–1.56]) and a weaker for Alzheimer’s disease (HRper SD decrease 1.07 [95% CI 5 1.03–1.13]). The intrabrother and twin analyses (taking shared family factors into account) showed attenuated risk estimates but with wide CIs. Discussion: Low early-life cognitive ability increases the risk of dementia before the age of 78 years. The association is partly explained by shared family factors. Ó 2017 the Alzheimer’s Association. Published by Elsevier Inc. All rights reserved.
Keywords:
Cohort study; Twin study; Cognitive ability; Dementia; Alzheimer’s disease; Vascular dementia
1. Introduction Dementia is a leading cause of disability, and with an increasing elderly population worldwide, the prevalence of dementia is expected to increase, despite reductions in incidence [1–4]. Alzheimer’s disease (AD) is the most common cause of dementia in the elderly, followed by vascular
The authors have declared that no conflict of interest exists. *Corresponding author. Tel.: 14538633780; Fax: 14538633977. E-mail address: [email protected]
dementia (VaD). The increasing awareness of shared modifiable risk factors such as cardiovascular risk factors [3] calls for an identification of such risk factors and effective strategies for prevention of dementia. In a life-course perspective, general early-life cognitive ability has been proposed as a potential risk factor for dementia. Low early-life cognitive ability reflects prenat al brain development [5–7] and is associated with subsequent morbidity and mortality [8–11]. This might increase the risk of dementia, which acts not only through cognitive reserve but also through poorer lifestyle and comorbidity, potentially influencing thrombotic and atherosclerotic processes and might
http://dx.doi.org/10.1016/j.jalz.2017.04.003 1552-5260/Ó 2017 the Alzheimer’s Association. Published by Elsevier Inc. All rights reserved. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
2
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
increase the risk of vascular lesions [6]. However, the influence of early-life cognitive ability and lifestyle on risk of developing AD and VaD is scarcely investigated, mainly due to the lack of access to information on early-life cognitive ability. In the study of 93 nuns from Milwaukee, USA, linguistic ability at the age of 22 years was associated with cognitive test scores and a diagnosis of AD at the age of 75 to 95 years [12]. In the Scottish 1921 birth cohort, 50 patients with dementia diagnosed after the age of 65 years had lower mental ability at the age of 11 years compared with controls, but this was not the case for 59 patients with early-onset dementia [13]. Furthermore, analysis of the Scottish birth cohort also showed that the 31 cases with VaD had lower early-life mental ability than matched controls, but there was no significant difference for 63 cases with AD [14]. More recently, two studies based on the Swedish Military Conscription Register [15,16] have shown that low cognitive ability in young adulthood was associated with dementia before the age of 60 years (dementia cases: n 5 487 [15] and n 5 657 [16]). Family studies have suggested that more than 50% of the between-subject variation in both early-life cognitive ability [17] and AD can be ascribed to genetic factors [17–19]. Environmental factors shared within families and attributed to cognitive ability include parental lifestyle and socioeconomic conditions [20]. Twin and sibling studies offer a unique design for controlling, in part, for shared genetic and familial factors [21,22] by taking advantage of the fact that twins and full siblings are almost always brought up together and that monozygotic (MZ) twins are perfectly matched genetically, whereas dizygotic (DZ) twins and full siblings share on average 50% of their segregating genes. To date, no studies have investigated the impact of shared familial factors on the association between earlylife cognitive ability and late-life dementia. The objectives of the study were to examine the association between cognitive ability in young adulthood and AD or VaD diagnosis and age of onset in men. We also wanted to assess whether this association could be replicated within brothers and male twin pairs. We hypothesize that low early-life cognitive ability increases the risk of dementia, and that familial factors including genetic constitution account for a considerable part of the expected association.
2. Methods 2.1. Study population The Danish Conscription Database (DCD) comprises 728,158 Danish men born between 1939 and 1959, who were conscripted for mandatory military service from 1957 until 1984 and were alive on April 1, 1968 when the Danish Civil Registration System was established. Through data linkage with the Danish Twin Registry and the Danish Civil Registration System, 14,408 male twins (4108 intact male twin pairs n 5 8216) and 81,002 brothers were
identified in the DCD. Of the 728,158 men, 63,172 (0.9 %) were excluded; 52,843 had no conscript examination data, whereas 5691, 1181, and 1458 had no information on cognitive ability, education, or height. Available for analysis were 666,986 men including 74,761 brothers and 7310 male twins (n 5 3655 pairs with information on zygosity for 1085 MZ and 2226 DZ pairs) with complete information on all data. The missing observations were mainly due to the lack of registry information for men who volunteered for military service or had been exempted from the draft board examination because of the medical conditions such as intellectual disability, psychiatric disorders, epilepsy, or type 1 diabetes. The percentage of missing observation was highest (8.2%) for men born after 1955 and lowest (5.9%) for those born before 1944. The DCD has been described in detail, previously [23]. All data linkages and analyses were approved by the Danish Data Protection Agency. 2.2. Assessment of cognitive ability Cognitive ability was assessed by the draft board intelligence test, the Børge Prien Prøve (BPP), which has been described previously [23,24] and used in other studies [9]. It is a group-administered paper and pencil test comprising four subtests (letter matrices, 19 items; verbal analogies, 24 items; number series, 17 items; and geometric figures, 18 items) to be completed within 45 minutes. This version of BPP has been used without changes since 1957. The numbers of correct answers in the four subtests are summed to a total score with a 0 to 78 range. The total BPP score has been shown to correlate with the full-scale Wechsler Adult Intelligence Scale (r 5 0.82) [25]. 2.3. Dementia outcome measures Information on any admission to a psychiatric or somatic ward from 1969 or 1977, respectively, until 2016 was obtained by linking the DCD to the Danish Psychiatric Central Registry and the Danish National Patient Registry using the person identification number as a key. These registers hold individual-level data on type of patient contact (inpatients, and, from 1995 onward, emergency and outpatients), diagnosis, and date of admission for all hospital admissions [26]. Mortality from all and specific causes of death was followed from April 1968 until April 2016 in the Danish Cause of Death Register. Hospital diagnosis and registered causes of death have been classified according to 8th revision of the International Classification of Disease (ICD-8) for the period 1969 to 1993, and according to the ICD-10 from 1994 onward. In the present study, the outcome of interest was the first hospital discharge or death from a main diagnosis of dementia (ICD-8: 290.00–290.99 and ICD-10: F00.0–F03.9; G30.0–G30.9) from 1969 until 2016. Dementia cases were subdivided into AD (ICD-8: 290.10 and ICD-10: F00.0–F00.9; G30.0–G30.9) and VaD (ICD-10: F01.0–F01.9) cases.
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
2.4. Validity of dementia diagnosis The validity of dementia diagnosis obtained from the previously mentioned registers has been assessed in two independent studies of patients admitted in 2003 [27] and 2008 [28]. They revealed that 70% [28] and 83% [27] of dementia cases diagnosed by an external rater according to ICD-10 confirmed the diagnosis recorded in the register [27,28] with the lowest validity in younger populations and for VaD. The validity of diagnosis based on ICD-8 has not been examined. In the DCD cohort, 663 and 5563 men were registered with a dementia diagnosis defined by either ICD-8 or ICD-10. 2.5. Covariables From the DCD, height, education, and any registered admission to a psychiatric hospital at study entry were included as covariables because these variables have been shown to be associated with early-life cognitive ability and risk of dementia [8,29]. Educational level reported by the men at conscription was coded into three categories comprising low (7th–9th grade), medium (vocational training or 10th–11th grade), and high (12th grade or more advanced) educational levels. Height was measured without shoes at the physical examination by a physician. 2.6. Statistical analyses
board examination (for men born 1951–1959 conscripted in 1969–1984; age 19 years). Follow-up was terminated at the age of first admission, death, emigration, or end of followup (April 1, 2016; age 57–78 years), whichever came first (Fig. 1). A few subjects (n 5 4) diagnosed with dementia before the start of follow-up were excluded. To examine the shape of the association between cognitive ability and dementia, we divided the BPP scores into six groups of equal size. This indicated a linear trend, but there were few outcomes in some strata. To obtain categories with sufficient number of cases and for comparability with previous studies [15], the BPP was grouped in tertiles. The BPP was also analyzed as a continuous variable. In multiple regression models, height, education, and any psychiatric hospital admission at study entry were included as covariables, both one at a time and together. Any interaction of cognitive ability with height, education, or birth cohort was examined comparing models with and without an interaction term using likelihood ratio tests, but there was no indication of significant multiplicative interaction. For siblings, we first examined how often the brother/co-twin with the lowest cognitive score was the first to be diagnosed with dementia compared with his higher scoring brother/co-twin (only including the first two brothers from families with more than two brothers). Secondly, we conducted a standard cohort analysis treating each twin or brother as individuals, taking the interdependence of observations within siblings into account by including a cluster term. In the intrasibling and twin analyses, a variable identifying membership of a twin pair or sibling group was included as a stratum variable, fixing the baseline hazard within siblings, but allowing it to differ between individuals who were not siblings. Thus, the hazard ratios (HRs) were estimated conditional on the family-level intercept. This corresponds to conditional logistic regression, and made it possible to
web 4C=FPO
The influence of cognitive ability in young adulthood on the incidence of dementia was analyzed using Cox regression with age as the underlying time scale. Person-years (py) of follow-up for a dementia diagnosis was accumulated from age on January 1, 1969 (for men born 1939–1950 conscripted in 1957–1968; age 19–30 years) or on the date of the draft
3
Fig. 1. Venn diagram with age and time for follow-up of the men born between 1939 and 1959 in the Danish Conscription Database. Abbreviations: ICD-8, 8th revision of the International Classification of Disease; ICD-10, 10th revision of the International Classification of Disease. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
4
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
compare siblings while keeping common factors fixed (i.e., family factors shared within siblings). We took subjects who died before they could have been admitted (competing risk) into account using Fine-Gray competing risk regression by the “stcrreg” command in STATA. This provides “subdistribution HRs,” which can be used as an alternative to Cox regression to determine whether the observed associations can be attributed to competing risks [30]. The proportional hazard assumption was examined by inspection of cumulative hazard plots. This showed that around the age of 60 years, the difference between curves for cognitive ability diminished, and, consequently, analyses for early- and late-onset dementia were split at that age. Subsequently, the test of proportional hazards revealed no violation of the assumptions. Finally, we examined the association between cognitive ability and dementia defined based on ICD-8 or ICD-10 in men born between 1939 and 1949 followed from the age of 45 to 65 years and in men born between 1950 and 1959 followed through the same ages. All analyses were carried out in STATA, version 12. 3. Results 3.1. Incidence of dementia diagnosed during follow-up The 666,986 men were followed for a median of 44.1 years (27.8 million py), and during that time, 6416 (0.96%) developed dementia (incidence rate 5 23.4 [95% confidence interval {CI} 5 22.8–23.9] per 100,000 py). Of these, 1760 (0.26%) and 970 (0.15%) were classified as AD or VaD, respectively, whereas 3686 dementia cases had an unspecified diagnosis (ICD-10: F03.9). 3.2. Cognitive ability and risk of dementia
3.3. Cognitive ability and risk of dementia in brothers and twins Among the 74,759 brothers and 7310 male twins, 340 (0.45%) and 60 (0.82%) developed dementia during follow-up, respectively. Supplementary Table 1 presents the intrasibling discordance in young adult cognitive ability for brothers and twins and shows that percentage of pairs with more than 1 standard deviation (SD) (.11 points difference) discordance on cognition was higher in brothers than in twins. In pairs (n 5 292) of brothers where one or both had developed dementia, the proportion of pairs in which the brother with the lowest score was diagnosed with
web 4C=FPO
Figs. 2–4 and Table 1 show the distribution of dementia diagnoses in relation to cognitive ability in young adulthood
and the corresponding crude and adjusted HRs. Fig. 2 shows that low cognitive ability in young adulthood was associated with an increased risk of dementia (HRlowest versus highest tertile 1.74 [95% CI 5 1.60–1.88] or HRper SD decrease 1.33 [95% CI 5 1.30–1.35]) with the strongest associations observed for VaD (Figs. 3 and 4). A total of 2538 (39.6%) of dementia cases were diagnosed before the age of 60 years, of these, 281 (15.0%) as AD and 298 (30.7%) as VaD, and cognitive ability seemed to be more strongly associated with the early-onset dementia (Fig. 2). Table 1 also shows that dementia was more frequent in men with low education or height, and both factors were associated with increased risk of dementia (HRlow versus high education 1.50 [95% CI 5 1.38–1.63] or HRlow versus high height 1.62 [95% CI 5 1.52–1.73]). However, adjustment for education and/ or height at conscription only slightly influenced the estimates (Figs. 2–4). Few men (0.3%) had been admitted to a psychiatric hospital at study entry, and no association with dementia risk was detected (HR 1.42 [95% CI 5 0.82– 2.44]) or influenced the association between cognitive ability and dementia. Taking competing mortality risk into account did not change the conclusions.
Fig. 2. Number with dementia and associations (HRs and 95% CIs) in relation to cognitive ability in young adulthood among 666,934 Danish men born between 1939 and 1959. Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
5
web 4C=FPO
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
Fig. 3. Number with Alzheimer’s disease and associations (HRs and 95% CIs) in relation to cognitive ability in young adulthood among 666,934 Danish men born between 1939 and 1959. Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation.
(HRper SD decrease 1.36[95% CI 5 1.08–1.73]) (Table 2), which was close to the estimates for the overall sample and for men with no brothers and those with no parental identification. The adjusted intrasibling analyses, which took into account shared family factors, indicated slightly attenuated risk estimates for brothers (HRper SD decrease 1.32 [95% CI 5 1.04–1.64]). In male twin pairs, the HRs were more attenuated but also imprecisely estimated
web 4C=FPO
dementia first was 57%. This percentage was higher than the 50% (assuming no difference) but was not related to the difference in cognitive ability within brother pairs (being 60%, 50%, and 59% in pairs with less than ½ SD, ½–1 SD, and more than 1 SD difference in cognitive scores, respectively). In the adjusted standard cohort analysis, we found a relation between cognitive ability and dementia in both brothers (HRper SD decrease 1.47 [95% CI 5 1.31–1.62]) and twins
Fig. 4. Number with vascular dementia and associations (HRs and 95% CIs) in relation to cognitive ability in young adulthood among 666,934 Danish men born between 1939 and 1959. Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
6
Table 1 The number (n) with dementia in relation to cognitive ability, education, and height in early adulthood among 666,934 Danish men born between 1939 and 1959 Total cohort
Total Cognitive ability Lowest Middle Highest Education Lowest Middle Highest Height Lowest Middle Highest Birth year 1939–1944 1945–1949 1950–1954 1955–1959 Prior psychiatric admission No Yes
Men with parental identification
Male twins
N (%)
Dementia n (%)
Alzheimer’s disease n (%)
Vascular dementia n (%)
N (%)
Dementia n (%)
N (%)
Dementia n (%)
666,986 (100)
6416 (0.96)
1760 (0.26)
970 (0.15)
245,998 (100)
1235 (0.50)
12,941
140 (1.08)
237,959 (35.68) 219,243 (32.87) 209,784 (31.45)
3142 (1.32) 1853 (0.85) 1421 (0.68)
742 (0.31) 518 (0.24) 500 (0.24)
404 (0.21) 297 (0.14) 169 (0.08)
71,859 (24.21) 84,961 (34.54) 89,178 (36.25)
587 (0.802) 360 (0.42) 288 (0.32)
5808 (44.86) 4009 (31.02) 3124 (24.12)
84 (1.45) 34 (0.85) 22 (0.70)
171,699 (25.74) 381,495 (57.20) 113,792 (17.06)
2042 (1.19) 3624 (0.95) 750 (0.66)
475 (0.28) 1015 (0.27) 270 (0.24)
346 (0.20) 525 (0.14) 99 (0.09)
54,001 (22.36) 135,709 (55.17) 55,288 (22.48)
404 (0.73) 644 (0.47) 187 (0.34)
3804 (29.41) 7350 (56.78) 1787 (13.81)
55 (1.45) 75 (1.02) 10 (0.56)
242,277 (36.32) 235,729 (35.34) 188,980 (28.33)
3012 (1.24) 2128 (0.90) 1276 (0.68)
781 (0.32) 585 (0.25) 389 (0.21)
495 (0.20) 325 (0.14) 150 (0.08)
72,720 (29.56) 87,852 (35.71) 85,426 (34.73)
508 (0.70) 405 (0.46) 322 (0.38)
5425 (41.92) 4442 (34.34) 3074 (23.84)
77 (1.42) 41 (0.92) 22 (0.72)
177,647 (26.63) 186,155 (27.91) 162,156 (24.31) 141,028 (21.15)
3385 (1.91) 1829 (0.98) 803 (0.50) 399 (0.28)
1112 (0.63) 441 (0.24) 143 (0.09) 64 (0.05)
540 (0.30) 270 (0.15) 101 (0.06) 52 (0.04)
15,601 (6.34) 23,069 (9.38) 75,303 (30.61) 132,025 (53.67)
286 (1.83) 240 (1.04) 352 (0.47) 357 (0.50)
3353 (25.91) 3810 (29.44) 3009 (23.31) 2769 (21.34)
81 (2.42) 34 (0.89) 18 (0.60) 7 (0.25)
666,048 (99.71) 1938 (0.29)
6403 (0.96) 13 (0.69)
1760 (0.26) 0 (0.00)
968 (0.14) 2 (0.10)
244,690 (99.47) 1308 (0.53)
1230 (0.50) 5 (0.38)
12,209 (99.70) 39 (0.30)
138 (1.07) 2 (5.13)
(HRper SD decrease 1.09 [95% CI 5 0.55–2.13]). When the twin analysis was made separately for MZ and DZ twin pairs, the estimates for MZ (HRper SD decrease 0.30 [95% CI 5 0.04–2.83]) and DZ (HRper SD decrease 1.09 [95% CI 5 0.50–2.32]) twin pairs became very imprecise as they were based on only 14 and 51 cases with dementia.
with overlapping CIs (HRICD-8: lowest versus highest tertile 3.64 [95% CI 5 2.78–4.76] and HRICD-10: lowest versus highest tertile 2.81 [95% CI 5 2.16–3.66]). Finally, we examined dementia outcomes in men with complete information as compared to those with partly and complete missing data. This showed that the latter group had higher risk of dementia (HR 1.70 [95% CI 5 1.51–1.91]), whereas there was no difference for those with missing data on single variables.
3.4. Supplementary analyses We also analyzed data including dementia coded as secondary diagnosis in the outcome definition. This showed the same patterns of associations as reported previously. When the analyses were repeated on diagnosis based on ICD-8 or ICD-10 in a subsample of men followed during the same age span, the estimates seemed to be somewhat stronger for diagnoses based on ICD-8 than ICD-10, but
4. Discussion 4.1. Main findings In this cohort of 666,986 men followed for a median of 44 years from about the age of 19 years, 1 SD decrease in cognitive ability measured at baseline was associated with a 33% higher risk of dementia. The risk estimates were
Table 2 Associations (adjusted hazard ratio and 95% CIs) between cognitive ability in young adulthood and subsequent risk of dementia among 666,934 Danish men born 1939–1959, showing results from standard cohort and intrasibling and twin analyses (168 and 34 affected pairs) Standard cohort analysis
Dementia cases/total number Cognitive ability Lowest Middle Highest Per SD increase
Intrasibling and twin analysis No parental information
No brother
Brothers
Twins
5181/420,988
895/171,237
340/74,761
69/7310
1.63 (1.50–1.77) 1.16 (1.07–1.26) 1 1.33 (1.28–1.37)
2.05 (1.67–2.53) 1.26 (1.06–1.54) 1 1.47 (1.30–1.61)
2.34 (1.65–3.31) 1.26 (0.97–1.78) 1 1.47 (1.31–1.62)
1.25 (0.53–2.91) 1.20 (0.52–2.80) 1 1.36 (1.08–1.73)
Abbreviations: CI, confidence interval; SD, standard deviation. FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
Brothers
Twins
1.89 (1.05–3.38) 1.11 (0.65–1.89) 1 1.32 (1.04–1.64)
1.71 (0.34–8.48) 2.69 (0.64–11.21) 1 1.09 (0.55–2.13)
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
stronger for VaD than AD and for early- versus late-onset dementia of all subtypes. The associations were slightly attenuated when adjusted for education and/or height measured at the age of 19 years. In the intrasibling and twin analysis, the associations between cognitive ability and dementia were attenuated indicating that family factors explain part of the association, although the estimates for brothers were imprecisely estimated, especially for the twins. 4.2. Comparison with other studies Our results are in accordance with previous studies on dementia with early [16] and late onset [14,16], but in contrast, the Scottish birth cohort showed no association between cognitive ability at the age of 11 years and early-onset dementia [13]. However, our study is comparable to the study of Swedish conscripts [16] reporting both an association between cognitive ability in early adulthood and early-onset dementia. The relations between cognitive ability in early life and AD and VaD were also examined in the Scottish birth cohort, and in accordance with our findings, the study showed an association between low cognitive ability at the age of 11 years and VaD. However, there was no relation with AD, possibly because the number of AD patients (n 5 63) was insufficient to detect a small difference in cognitive ability between patients and controls. In the present study, education in young adulthood had only minor influence on the association between cognitive ability and dementia outcomes. Similarly, in the Nun Study [12], linguistic ability at the age of 22 years was associated with cognitive ability 58 years later even when adjusted for education. In the Swedish conscript cohort, adjustment for education at the age of 18 years had the largest, although still modest, attenuating influence on the association between cognitive ability and dementia, whereas parental education and cardiovascular fitness did not influence the risk estimates [16]. The Swedish researchers also explored the influence of neurological and psychiatric disorders emerging up to 15 years after conscription on the risk estimates. Although the exclusion of subjects with these disorders did not seem to influence the association between cognitive ability and dementia, it is not clear how the time dependency of these comorbidities was accounted for in the analyses. In our study, adjustment for prior psychiatric hospital admission had no influence on the association between cognitive ability and dementia. Our study is the first to suggest that family environment and genetic constitution might explain part of the association, but risk estimates were not very precise as they were based on the subsample of 74,761 brothers and 7310 male twins with 340 and 69 dementia cases, respectively. The information on parental identity was obtained from the Danish Civil Registration System and is almost complete for persons born after 1955, but the availability decreases rapidly for those born earlier from about 10% for those born in 1950 to less than 1% for those born before 1940
7
[31]. Consequently, we had missing information on parents for more than two-thirds of cohort members, but sensitivity analyses showed no difference in the pattern of associations between individuals with and without parental information. Some (3.3%) of the brothers represented a twin pair, but since the two populations of brothers and twins were sampled from the Danish Civil Registration System and the Danish Twin Registry, with different degrees of completeness, we decided to analyze two independent substudies. 4.3. Study strength and limitations The strength of the present analysis is that it was based on a relatively large sample compared with the Scottish studies [13,14] and that the sample covered older birth cohorts (1939–1959) than the Swedish conscription cohort (1950–1987) [16]. Furthermore, cognitive ability was assessed in young adulthood where it can be expected to be relatively stable. In our study, outcome information was retrieved from national registers with sensitivity measures .70% [26–28]. The diagnosis of dementia in young persons might be less valid because of a different prevalence of subtypes and symptom profile and a range of mainly psychiatric or neurological conditions that may be misdiagnosed as dementia. Thus, we cannot exclude that the observed, stronger association between cognitive ability and dementia diagnosed before the age of 60 years could reflect that low cognitive ability was misinterpreted as dementia in younger persons. On the other hand, information on cognitive ability was not available for those exempted from conscription because of mental disability and severe psychiatric disorder, and our sensitivity analysis showed that subjects with missing data had higher risk of dementia than those available for the present study. This suggests that our estimates of the relation between cognitive ability and dementia might not be explained by confounding from such disorders. Our study covered 20 birth cohorts followed for outcomes for a median of 44 years from ages 18 to 30 years. Thus, follow-up ended when cohort members were between 56 and 76 years and, consequently, our study does not include old-age dementia outcomes. During this time period, ICD coding has changed from ICD-8 to ICD-10, and the coverage of the hospital registers has increased because of the inclusion of outpatient contacts from 1995. However, we did not find any major difference in the risk estimates for the association between cognitive ability and earlyonset dementia across birth cohorts or in substudies with dementia defined based on ICD-8 and ICD-10, respectively. Dementia is more prevalent in women, and agedependent differences in the prevalence between men and women have been reported [32]. Our cohort consisted of men only, and the results are not necessarily generalizable to women. However, results from the Nun Study and the Scottish birth cohorts support a relationship between
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
8
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
cognitive ability in the young and risk of dementia also in women. Although the study is large, the analyses, especially those on siblings and twins, are still affected by lack of statistical power.
4.4. Possible mechanism and implications of our findings Mechanisms linking cognitive ability in early life with later health outcomes [5,14,19] include confounding from genetic constitution, familial factors, and poor health during early childhood or influences mediated through poor lifestyle and health. In our study, education, height, and early psychiatric comorbidity, which are closely related to deprivation in early life, were not important confounders, suggesting that attenuation of the association in the intrasibling and twin analyses should rather be attributed to other family factors, such as genetic and environmental factors contributing to cognitive reserve as a compensatory mechanism for coping with brain pathology [33]. The relatively strong association between early-life cognitive ability and VaD suggests a combined risk of limited cognitive reserve and poor lifestyle and health. Studies report on an association of early-life cognitive abilities with risk of cardiovascular events [34] and cardiovascular risk factors are not only associated with an increased risk of VaD but also AD [35]. The Flynn effect, that is, an observation of secular increases in young adult cognitive scores across a range of populations, is strong evidence for an important environmental effect, including lifestyle. This effect has occurred far too quickly to be explained purely by changes in the gene pool or other biological effects [23,36], but it may have contributed to the recent decline in the incidence of dementia observed in Western populations [2,3]. In conclusion, low cognitive ability in young adulthood seems to be associated with higher risk of all types of dementia. Consequently, early-life cognitive ability may provide a further modifiable risk factor for dementia. Thus, we suggest that prevention of dementia should be initiated early in life and attempts should be made to preserve cognitive ability especially among those with low ability.
Acknowledgments The work was supported by the Danish Medical Research Council [grant number 09-063599 and 09-069151] and The Velux Foundations [grant number 31205].
Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jalz.2017.04.003.
RESEARCH IN CONTEXT
1. Systematic review: In a study of 93 American nuns, linguistic ability at the age of 22 years was associated with late-life Alzheimer’s disease (AD). Later, the Scottish birth cohort showed that low mental ability at the age of 11 years increased the risk of vascular dementia (VaD) but not of AD. Recently, Swedish conscription data have shown that low cognitive ability was associated with early-onset dementia. We are the first to examine whether an association between early-life cognitive ability and dementia is replicated within brothers and male twins. 2. Interpretation: In accordance with the previous studies, low cognitive ability was associated with increased risk of dementia before the age of 75 years. The novel findings were the stronger association for VaD than for AD, and the contribution of environmental factors shared within families to the association between low early-life cognitive ability and dementia. 3. Future direction: Future research should further explore the factors mediating the association between cognitive ability and dementia and any differential influence of familial factors on the risk of VaD and AD.
References [1] Wu YT, Fratiglioni L, Matthews FE, Lobo A, Breteler MM, Skoog I, et al. Dementia in Western Europe: epidemiological evidence and implications for policy making. Lancet Neurol 2016;15:116–24. [2] Matthews FE, Stephan BC, Robinson L, Jagger C, Barnes LE, Arthur A, et al. A two decade dementia incidence comparison from the Cognitive Function and Ageing Studies I and II. Nat Commun 2016;7:11398. [3] Satizabal CL, Beiser AS, Chouraki V, Chene G, Dufouil C, Seshadri S. Incidence of Dementia over three decades in the Framingham Heart Study. N Engl J Med 2016;374:523–32. [4] Wimo A, Guerchet M, Ali GC, Wu YT, Prina AM, Winblad B, et al. The worldwide costs of dementia 2015 and comparisons with 2010. Alzheimers Dement 2017;13:1–7. [5] Papachristou E, Frangou S, Reichenberg A. Expanding conceptual frameworks: life course risk modelling for mental disorders. Psychiatry Res 2013;206:140–5. [6] Richards M, Deary IJ. A life course approach to cognitive reserve: a model for cognitive aging and development? Ann Neurol 2005; 58:617–22. [7] Lynch J, Smith GD. A life course approach to chronic disease epidemiology. Annu Rev Public Health 2005;26:1–35. [8] Osler M, Lawlor DA, Nordentoft M. Cognitive function in childhood and early adulthood and hospital admission for schizophrenia and
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce
M. Osler et al. / Alzheimer’s & Dementia - (2017) 1-9
[9]
[10]
[11] [12]
[13] [14]
[15]
[16]
[17]
[18] [19]
[20]
[21]
[22]
bipolar disorders in Danish men born in 1953. Schizophr Res 2007; 92:132–41. Christensen GT, Mortensen EL, Christensen K, Osler M. Intelligence in young adulthood and cause-specific mortality in the Danish Conscription Database – a cohort study of 728,160 men. Intelligence 2016;59:64–71. Calvin CM, Deary IJ, Fenton C, Roberts BA, Der G, Leckenby N, et al. Intelligence in youth and all-cause-mortality: systematic review with meta-analysis. Int J Epidemiol 2011;40:626–44. Wraw C, Deary IJ, Gale CR, Der G. Intelligence in youth and health at age 50. Intelligence 2015;53:23–32. Snowdon DA, Kemper SJ, Mortimer JA, Greiner LH, Wekstein DR, Markesbery WR. Linguistic ability in early life and cognitive function and Alzheimer’s disease in late life. Findings from the Nun Study. JAMA 1996;275:528–32. Whalley LJ, Starr JM, Athawes R, Hunter D, Pattie A, Deary IJ. Childhood mental ability and dementia. Neurology 2000;55:1455–9. McGurn B, Deary IJ, Starr JM. Childhood cognitive ability and risk of late-onset Alzheimer and vascular dementia. Neurology 2008; 71:1051–6. Nordstrom P, Nordstrom A, Eriksson M, Wahlund LO, Gustafson Y. Risk factors in late adolescence for young-onset dementia in men: a nationwide cohort study. JAMA Intern Med 2013;173:1612–8. Nyberg J, Aberg MA, Schioler L, Nilsson M, Wallin A, Toren K, et al. Cardiovascular and cognitive fitness at age 18 and risk of early-onset dementia. Brain 2014;137:1514–23. Polderman TJ, Benyamin B, de Leeuw CA, Sullivan PF, van Bochoven A, Visscher PM, et al. Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nat Genet 2015; 47:702–9. Deary IJ, Johnson W, Houlihan LM. Genetic foundations of human intelligence. Hum Genet 2009;126:215–32. Gatz M, Reynolds CA, Fratiglioni L, Johansson B, Mortimer JA, Berg S, et al. Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry 2006;63:168–74. Nisbett RE, Aronson J, Blair C, Dickens W, Flynn J, Halpern DF, et al. Intelligence: new findings and theoretical developments. Am Psychol 2012;67:130–59. McGue M, Osler M, Christensen K. Causal inference and observational research: the utility of twins. Perspect Psychol Sci 2010; 5:546–56. D’Onofrio BM, Class QA, Rickert ME, Sujan AC, Larsson H, KujaHalkola R, et al. Translational epidemiologic approaches to understanding the consequences of early-life exposures. Behav Genet 2016;46:315–28.
9
[23] Christensen GT, Molbo D, Angquist LH, Mortensen EL, Christensen K, Sorensen TI, et al. Cohort profile: The Danish Conscription Database(DCD): a cohort of 728,160 men born from 1939 through 1959. Int J Epidemiol 2015;44:432–40. [24] Teasdale TW. The Danish draft board’s intelligence test, Borge Priens Prove: psychometric properties and research applications through 50 years. Scand J Psychol 2009;50:633–8. [25] Mortensen EL, Reinisch JM, Teasdale TW. Intelligence as measured by the WAIS and a military draft board group test. Scand J Psychol 1989;30:315–8. [26] Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sorensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7:449–90. [27] Phung TK, Andersen BB, Hogh P, Kessing LV, Mortensen PB, Waldemar G. Validity of dementia diagnoses in the Danish hospital registers. Dement Geriatr Cogn Disord 2007;24:220–8. [28] Salem LC, Andersen BB, Nielsen TR, Stokholm J, Jorgensen MB, Rasmussen MH, et al. Overdiagnosis of dementia in young patients a nationwide register-based study. Dement Geriatr Cogn Disord 2012;34:292–9. [29] Vagero D, Modin B. Commentary: the associations between height, cognition, and education and their relevance for health studies. Int J Epidemiol 2006;35:663–4. [30] Andersen PK, Geskus RB, de Witte T, Putter H. Competing risks in epidemiology: possibilities and pitfalls. Int J Epidemiol 2012; 41:861–70. [31] Pedersen CB, Gotzsche H, Moller JO, Mortensen PB. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 2006;53:441–9. [32] Lobo A, Launer LJ, Fratiglioni L, Andersen K, Di Carlo A, Breteler MM, et al. Prevalence of dementia and major subtypes in Europe: a collaborative study of population-based cohorts. Neurologic Diseases in the elderly research group. Neurology 2000;54:S4–9. [33] Stern Y. Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 2012;11:1006–12. [34] Dobson KG, Chow CH, Morrison KM, Van Lieshout RJ. Associations between childhood cognition and cardiovascular events in adulthood: a systematic review and meta-analysis. Can J Cardiol 2017;33:232–42. [35] Barnes DE, Yaffe K. The projected effect of risk factor reduction on Alzheimer’s disease prevalence. Lancet Neurol 2011;10:819–28. [36] Trahan LH, Stuebing KK, Fletcher JM, Hiscock M. The Flynn effect: a meta-analysis. Psychol Bull 2014;140:1332–60.
FLA 5.4.0 DTD JALZ2388_proof 24 May 2017 10:35 am ce