Facets of Personality and Risk of Cognitive Impairment: Longitudinal Findings in a Rural Community from Sardinia

Abstract

Background:

Few studies have examined the associations between personality facets and dementia risk and rarely included individuals from rural settings or with low education.

Objective:

To examine the association between personality and the risk of cognitive impairment.

Methods:

Participants (N = 1,668; age 50 to 94 at baseline; 56.4% women; 86.5% less than high school diploma) were from a rural region of Sardinia (Italy) who completed the Revised NEO Personality Inventory (NEO-PI-R) during the first wave (2001–2004) and the Mini-Mental State Examination (MMSE) at waves two to five (2005–2021). Cox regression was used to test personality and covariates as predictors of cognitive impairment based on MMSE education-adjusted cutoffs.

Results:

During the up to 18-year follow-up (M = 10.38; SD = 4.76), 187 individuals (11.2%) scored as cognitively impaired. Participants with higher neuroticism (particularly the depression facet [HR = 1.22, 95% CI = 1.06–1.40]), and lower agreeableness (particularly the modesty facet [HR = 0.83, 95% CI = 0.71–0.97]) and lower conscientiousness (particularly the dutifulness facet [HR = 0.78, 95% CI = 0.67–0.92]) were at higher risk of cognitive impairment. Lower warmth ([HR = 0.75, 95% CI = 0.65–0.87], facet of extraversion) and ideas ([HR = 0.76, 95% CI = 0.65–0.89], facet of openness) were also associated with increased risk of impairment. These associations were virtually unchanged in models that accounted for other risk factors, including smoking, depression, obesity, hypertension, diabetes, and apolipoprotein E (APOE) ɛ4 carrier status. Across the five domains, sex and the APOE variant did not moderate the associations.

Conclusion:

In a sample with demographic characteristics underrepresented in dementia research, this study identifies personality domains and facets most relevant to the risk of cognitive impairment.

Keywords

Cognition dementia longitudinal study personality risk factors

INTRODUCTION

Research on clinical and behavioral risk factors for dementia can help identify targets of intervention for dementia prevention. Among the psychological risk factors for cognitive impairment, a growing body of research has found that personality traits are implicated in cognitive health [1]. Personality traits are defined as individual differences in enduring patterns of feeling, thinking, and behaving [2]. These traits are prominent in lifespan theories of aging [3, 4] for their relevance to crucial life outcomes, such as well-being, social connectedness, economic resources, and longevity [5]. Individual differences in personality are also related to health behaviors and the risk of common diseases, including neurodegenerative conditions like Parkinson’s [6] and Alzheimer’s [7 –12] diseases. Indeed, a recent meta-analysis (> 30,000 participants) found that higher neuroticism (a measure of negative emotionality and vulnerability to stress) and lower conscientiousness (the tendency to be organized, disciplined, and responsible) are robust predictors of dementia risk [1]. Findings were more mixed for the other three major domains of personality, but the meta-analysis suggested higher openness (the tendency to be imaginative and prefer variety), extraversion (a measure of sociability and positive emotionality), and agreeableness (the tendency to be trusting, cooperative, and altruistic) have small protective effects. The meta-analysis [1] further reported that the effects were highly consistent across samples, with only one significant moderator: The country of study participants, with associations stronger for extraversion and weaker for agreeableness among European compared to US samples. Of note, the associations were slightly stronger in studies that used the “gold standard” clinical diagnosis, but still significant in studies that used cognitive screening tests. The use of screening instruments to ascertain cognitive impairment is likely to underestimate associations but facilitate the study of larger, more diverse, and more representative samples.

The scope of this study was to examine whether broad traits and narrow facets of personality are prospectively associated with the risk of cognitive impairment in a sample from four towns in rural Sardinia— an Italian island in the Mediterranean Sea. This study addresses at least two significant questions.

First, it investigates the associations in a sample from a rural population with a broad range of educational levels. Such demographic characteristics are relatively rare in dementia studies [13], as well as in personality research [1, 14], which generally rely on samples with an overrepresentation of individuals from urban/suburban settings and with higher levels of education. Some research suggests that rural residence is associated with a higher risk of dementia [15]. Similarly, low education is a risk factor for dementia [16, 17]. Among other noteworthy features, this sample is from a founder population with a relatively low level of genetic admixture [18], which reduces potential confounding due to genetic admixture. The apolipoprotein E (APOE) ɛ4 allele frequency is estimated to be about 6% in the Sardinian population [19, 20], which is relatively lower than the 14% in other European-ancestry samples [21]. This region of Sardinia is also known as a blue zone for the relatively high proportion of centenarians [22]. Given the sample characteristics, the study addresses the question of whether the personality associations with cognitive impairment replicate in communities underrepresented in previous research.

Second, the study aims to advance the field with an in-depth assessment of personality using a well-established measure of 30 facets, six for each of the five major domains of personality. The facets are narrower traits hierarchically ordered under the five major domains. Some evidence suggests that lower-order traits, like the facets, could have stronger associations than the broader domains [23]. The analysis at the level of personality facets provides an in-depth understanding of which personality facet drives the associations with cognitive impairment. It is possible that only one or a few facets within a domain are associated with the outcome. It is also possible that some facets within a domain could have associations in opposite directions, which could lead to null effects at the domain level [24]. To our knowledge, only three previous studies have examined the association between facets of personality and risk of impairment or dementia. Of these three studies, only one (N = 1,671) [25] examined facets for all five domains. One study examined the six facets of neuroticism (N = 785) [26], and one focused on six facets of conscientiousness (N = 11,181) [27]. We and others have argued for potentially higher predictive power of facets [23 –28], but questions remain on the replicability of such findings.

The current study explored the associations between personality traits and cognitive performance (overall and domain scores) and tested personality traits as predictors of cognitive impairment at follow-up. Based on meta-analytic findings [1], we hypothesize that higher neuroticism and lower conscientiousness will be associated with higher risk of cognitive impairment. We do not have strong hypotheses for the other broad domains, but higher scores on extraversion, openness, and agreeableness could be protective. We further test whether the associations are independent of other major clinical, behavioral, and genetic risk factors (e.g., APOE). These other risk factors could confound or mediate the association between personality and cognitive impairment. However, based on past research [25 –27], we expect the associations of personality traits to be only slightly attenuated when accounting for other risk factors. We further test whether the associations are moderated by age, sex, and APOE ɛ4 carrier status. The analyses at the facet level can point to which specific aspect of the broad domains is responsible for the associations. We compare the findings to previous work [24] to test whether the associations at the facet level are replicable.

METHODS

Participants

Participants were part of the SardiNIA project, an ongoing genetic study of age-related conditions [18]. The sample recruitment and characteristics have been described in detail elsewhere [18, 29]. Briefly, about 62% of the eligible population of four adjacent towns in a rural region of Sardinia (Italy) were recruited between 2001 and 2004 [18]. Consistent with the idea of a founder population, almost all participants were native-born of the region, and at least 96% were known to have all grandparents born in the same province. Participants were assessed by trained staff in a building dedicated to the study. All participants signed informed consent to study protocols approved by the Sardinian Regional Ethics Committee (protocol no. 2171/CE). The study is ongoing, and participants have been assessed every three to four years, up to five waves.

Personality was assessed at wave 1. The cognitive screening measure was administered at waves 2 to 5, but only to participants aged ≥65 at waves 4–5. We restricted the analyses to participants with age ≥50 at wave 1 who were eligible for the cognitive screening at most follow-ups. See Fig. 1 for the selection of participants included in the analyses. Of the 6,162 individuals recruited at baseline [29], 5,669 participants had valid personality data, of whom 1,909 were aged 50+ at wave 1. Of these participants, 241 participants did not have follow-up cognitive screening, while 1,668 had at least one cognitive assessment at follow-up and thus could be included in the analyses. Attrition analyses found that compared to participants who had at least one follow-up assessment, participants without follow-up data were significantly older and had lower education (p < 0.01), but there were no significant differences for sex or any of the five major personality domains (p > 0.05).

Fig. 1

Flow chart of study participants selection and cognitive impairment outcome.

Measures: Personality

Participants completed the Italian version [30] of the Revised NEO Personality Inventory (NEO-PI R) [2]. The questionnaire consists of 240 items that assess 30 facets, six for each of the five domains. The items were answered on a 5-point Likert scale, from “strongly disagree” to “strongly agree”. Data across languages and cultures support the reliability and validity of the NEO-PI R [31]. Indeed, in the SardiNIA cohort, the factor structure showed high congruence with the normative structure (Tucker phi ≥0.91) and high internal consistency (Cronbach alpha ≥0.80) for the five factors [29]. A large literature further supports the validity of the scales across age groups, clinical and non-clinical samples, and self-report and observer rating methods [2, 31].

Participants completed the self-reported version of the NEO-PI R on paper, or a trained tester read each item and reported the answers that the participant provided [29]. Participants that required assistance were generally older and with lower levels of education. We accounted for this difference in assessment method in all analyses.

Measures: Cognition

The Mini-Mental State Examination (MMSE) [32] is widely used in research and clinical settings as a screening instrument for cognitive status in adults. The MMSE assesses several areas of cognitive function: orientation to time (5 points) and place (5), three-word registration (3), attention and calculation (counting backward by seven or spelling backward; 5), three-word recall (3), language (naming and comprehension; 8), and visual construction (copy of intersecting pentagons; 1). The total score can range from 0 to 30, and a score < 24 is often used as the cutoff for likely cognitive impairment or dementia [32]. Given that cognitive performance varies by educational level, we used Kochhann and colleagues’ [33] education-adjusted cutoffs: up to elementary school, MMSE < 21; junior high, MMSE < 22; high school, MMSE < 23; and university degree, MMSE < 24. The MMSE has been validated and standardized in several languages, including Italian [34]. The MMSE was administered by a clinician.

Measures: Covariates

The covariates included participants’ reported age in years, sex (female or male), years of education (from 0 to 17), and current smoking status (Do you smoke? No/Yes). As part of the medical history, participants were also asked about physician diagnosis of hypertension, diabetes, and depression (e.g., Has a doctor ever told you have had high blood pressure? No/Yes for each condition). The covariates also included obesity (body mass index ≥30) based on staff assessed weight and height and the APOE ɛ4 carrier status based on rs429358 and rs7412 (carrier versus not a carrier). The literature supports each covariate as a risk/protective factor for dementia [16, 17].

Statistical analyses

We calculated descriptive statistics for the study variables as means and standard deviations (SD) or proportions. For descriptive purposes, we used partial correlations accounting for age, sex, and personality assessment method (self-reported versus interview) to examine the association between the personality traits (wave 1) and the continuous MMSE total and cognitive domain scores (wave 2). For the primary analyses, we conducted Cox regression to examine the association between personality and risk of cognitive impairment. For participants who became impaired, time was computed as the difference between the age of personality assessment and the age at MMSE below the impairment cutoff. Participants who remained unimpaired were censored at their last MMSE evaluation and time was computed as the difference between the personality and the last MMSE assessment. Covariates in Model 1 were age, sex, education, and personality assessment method. Model 2 was Model 1 covariates and smoking, hypertension, diabetes, depression, obesity, and APOE ɛ4 carrier status. The proportional hazard assumption was met (ps > 0.01). The analyses were conducted separately for each personality domain and facet. Personality scores were z-scored (M = 0; SD = 1) so that hazard ratios (HRs) and 95% confidence intervals corresponded to a 1 SD difference. Moderation was tested by adding the interaction between each of the five domains and sex, age, and APOE ɛ4 carrier status. Each interaction was tested separately, and sex, age, education, and personality assessment method were included as covariates in all moderation analyses. We did not use correction for multiple testing to reduce the risk of false negatives [35] and because we tested evidence-based hypotheses for the five domains. For the facets, we examined whether the pattern of associations was due to chance (type I error) by correlating the effect sizes from this sample with those of a previous study based on the Baltimore Longitudinal Study of Aging (BLSA) [25]. Such approach reduces the risk of false negatives and focuses on the replicability of findings [36].

RESULTS

Table 1 presents descriptive statistics. At baseline, participants’ age ranged from 50 to 94, 56.4% were women, and 86.5% had less than a high school diploma. Table 2 presents partial correlation coefficients between personality traits at wave 1 and the domains and total MMSE scores at wave 2. Neuroticism had a negative association and Openness, Agreeableness, and Conscientiousness had a positive association with the total MMSE scores. Similar associations were found for the MMSE domains, especially for orientation to place, three-word recall, attention, and language.

Table 1

Descriptive statistics

	Min	Max	Mean Or N	SD Or %
Age baseline (y)	50.00	93.70	61.48	8.00
Age impaired^a (y)	53.30	99.20	74.39	8.25
Time^b (y)	2.50	17.90	10.38	4.76
Time^b (not impaired) (y)	2.50	17.90	10.93	4.58
Time^b (impaired) (y)	2.50	17.30	6.02	3.83
Neuroticism	27	153	89.68	17.67
Extraversion	49	156	102.19	13.88
Openness	50	157	96.13	14.95
Agreeableness	70	175	124.00	14.51
Conscientiousness	63	173	124.50	14.45
Assessment method^c	0	1	509	30.5%
Education	0	17	6.57	3.76
No degree (< 5)			179	10.7%
Elementary (5)			766	45.9%
Junior high (8)			498	29.9%
High school (13)			172	10.3%
University degree (≥17)			53	3.2%
Sex (female)	0	1	941	56.4%
Smoker	0	1	145	8.7%
Obesity	0	1	481	28.8%
Diabetes	0	1	110	6.6%
Hypertension	0	1	519	31.1%
Depression	0	1	134	8.0%
APOE ɛ4 alleles	0	2	243	7.3%
APOE ɛ4 carrier	0	1	239	14.3%
Wave 2 MMSE (n = 1,603)	11	30	26.36	3.10
Wave 3 MMSE (n = 1,121)	13	30	26.51	3.08
Wave 4 MMSE (n = 757)	5	30	27.31	3.16
Wave 5 MMSE (n = 756)	9	30	27.44	2.95

N = 1,668 or indicated in the Table. MMSE, Mini-Mental State Examination. ^aAge impaired is the age at interview visit for the first time below the MMSE cutoff. ^bThe variable time is the years between the personality assessment and the last MMSE (for those censored as not impaired) or the first MMSE with scores below the cutoff for those who became impaired. ^cAssessment method: personality assessed via interview = 1 or self-report = 0.

Table 2

Partial correlations between personality and domains and total MMSE scores

	Time	Place	Registration	Recall	Attention	Language	Spatial	Total
Neuroticism	–0.03	–0.09	–0.02	–0.06	–0.09	–0.05	–0.02	–0.12
Extraversion	–0.03	0.00	0.03	0.00	0.03	0.00	0.00	0.01
Openness	0.01	0.11	0.03	0.07	0.12	0.06	0.00	0.14
Agreeableness	0.06	0.11	0.01	0.04	0.05	0.11	0.05	0.13
Conscientiousness	0.04	0.09	0.02	0.03	0.05	0.06	0.02	0.09
N1: Anxiety	0.00	–0.04	–0.02	–0.02	–0.05	–0.01	–0.02	–0.05
N2: Angry hostility	–0.02	–0.08	–0.05	–0.03	–0.04	–0.06	–0.01	–0.08
N3: Depression	–0.04	–0.11	0.04	–0.04	–0.11	–0.04	–0.02	–0.13
N4: Self-consciousness	–0.01	–0.07	0.00	–0.05	–0.06	–0.03	0.01	–0.08
N5: Impulsiveness	–0.05	–0.05	–0.04	–0.05	–0.05	–0.05	0.00	–0.08
N6: Vulnerability	0.00	–0.04	–0.01	–0.06	–0.04	–0.03	–0.03	–0.06
E1: Warmth	–0.02	0.07	0.05	0.02	0.03	0.07	0.00	0.06
E2: Gregariousness	0.01	0.00	0.05	–0.04	0.01	0.00	0.03	0.00
E3: Assertiveness	0.02	–0.01	0.03	0.03	0.08	0.02	0.00	0.06
E4: Activity	0.00	0.01	0.02	0.01	0.02	–0.02	0.01	0.01
E5: Excitement-seeking	–0.07	–0.08	0.00	–0.02	–0.04	–0.08	–0.03	–0.10
E6: Positive emotions	–0.03	0.01	–0.03	0.00	0.00	0.02	0.00	0.00
O1: Fantasy	–0.03	0.06	0.01	0.05	0.04	0.03	–0.02	0.05
O2: Aesthetics	0.01	0.12	0.03	0.01	0.05	0.05	0.00	0.09
O3: Feelings	0.06	0.05	0.00	0.06	0.12	0.04	0.02	0.13
O4: Actions	–0.02	0.04	0.02	0.02	0.04	0.02	–0.01	0.04
O5: Ideas	0.02	0.12	0.04	0.06	0.15	0.06	0.01	0.16
O6: Values	0.01	0.00	–0.02	0.06	0.03	0.00	–0.01	0.03
A1: Trust	0.03	0.09	0.02	0.04	0.02	0.04	0.04	0.08
A2: Straightforwardness	0.05	0.10	–0.03	0.05	0.07	0.11	0.04	0.13
A3: Altruism	0.02	0.04	0.03	0.00	0.00	0.03	0.04	0.03
A4: Compliance	0.06	0.04	0.01	0.00	–0.01	0.02	0.01	0.03
A5: Modesty	0.04	0.05	–0.03	0.04	0.04	0.06	0.01	0.08
A6: Tender-mindedness	0.00	0.08	0.04	0.02	0.04	0.11	0.04	0.09
C1: Competence	0.03	0.06	0.01	0.01	0.03	0.02	0.01	0.05
C2: Order	0.01	0.05	0.06	–0.01	0.03	0.02	–0.02	0.04
C3: Dutifulness	0.03	0.06	–0.02	0.06	0.05	0.08	0.02	0.10
C4: Achievement striving	0.00	0.05	0.00	0.02	0.02	–0.01	0.00	0.03
C5: Self-discipline	0.06	0.10	0.01	0.05	0.08	0.09	0.03	0.13
C6: Deliberation	0.01	0.03	0.03	0.02	0.00	0.03	0.02	0.03

N = 1,603. Partial correlations accounting for age, sex, and personality assessment method. Bold p < 0.05.

Of the 1,668 participants, 187 (11%) scored in the impaired range over the follow-up that spanned up to 18 years (M = 10.37; SD = 4.76) and a total of 17,306 person-years. Results of the Cox regression (separately for each personality domain and facet) are in Table 3. Higher neuroticism and lower conscientiousness were risk factors for scoring in the cognitive impairment range. We also found that higher openness and agreeableness were protective. Effect sizes were similar across the four domains (neuroticism, conscientiousness, openness, and agreeableness), with a roughly 20% difference in risk for every 1 SD difference in each domain. Extraversion was not a significant predictor. Model 2 indicated that findings were robust when accounting for other common risk factors for dementia, and there was little attenuation of the personality associations (see Supplementary Table 1 for coefficients for all predictors). The small differences between Model 1 and 2 results indicate that the associations between personality traits and risk of impairment were mostly independent of clinical, behavioral, and genetic risk factors.

Table 3

Results from separate Cox regression with personality factors and facets as predictors of MMSE scores below the impaired cutoff

	Model 1	95% CI			Model 2	95% CI
Traits	HR	Low	High	p	HR	Low	High	p
Neuroticism	1.21	1.06	1.39	0.005	1.19	1.04	1.37	0.015
Extraversion	0.92	0.80	1.06	0.272	0.94	0.82	1.08	0.389
Openness	0.81	0.69	0.94	0.006	0.81	0.69	0.95	0.008
Agreeableness	0.82	0.70	0.96	0.013	0.82	0.70	0.96	0.015
Conscientiousness	0.82	0.70	0.97	0.017	0.83	0.70	0.98	0.024
N1: Anxiety	1.08	0.94	1.25	0.282	1.06	0.91	1.22	0.464
N2: Angry hostility	1.10	0.97	1.26	0.150	1.09	0.95	1.25	0.205
N3: Depression	1.22	1.06	1.40	0.005	1.20	1.04	1.38	0.012
N4: Self-consciousness	1.17	1.02	1.34	0.022	1.16	1.01	1.34	0.031
N5: Impulsiveness	1.21	1.04	1.40	0.011	1.19	1.03	1.38	0.020
N6: Vulnerability	1.14	0.99	1.31	0.075	1.11	0.96	1.28	0.168
E1: Warmth	0.75	0.65	0.87	< 0.001	0.76	0.66	0.89	< 0.001
E2: Gregariousness	0.97	0.84	1.11	0.630	0.98	0.85	1.12	0.757
E3: Assertiveness	0.92	0.80	1.06	0.268	0.93	0.80	1.07	0.303
E4: Activity	0.96	0.83	1.11	0.552	0.97	0.84	1.12	0.650
E5: Excitement-seeking	1.14	0.98	1.32	0.089	1.14	0.98	1.32	0.090
E6: Positive emotions	0.97	0.85	1.11	0.662	0.99	0.86	1.13	0.865
O1: Fantasy	0.97	0.84	1.13	0.714	0.97	0.84	1.13	0.703
O2: Aesthetics	0.88	0.76	1.01	0.072	0.89	0.77	1.03	0.106
O3: Feelings	0.90	0.78	1.05	0.190	0.89	0.77	1.04	0.143
O4: Actions	0.88	0.76	1.02	0.093	0.88	0.76	1.02	0.097
O5: Ideas	0.76	0.65	0.89	< 0.001	0.76	0.65	0.90	< 0.001
O6: Values	0.93	0.80	1.08	0.324	0.93	0.80	1.09	0.369
A1: Trust	1.00	0.87	1.15	0.975	1.02	0.89	1.16	0.823
A2: Straightforwardness	0.83	0.71	0.97	0.019	0.83	0.71	0.97	0.021
A3: Altruism	0.85	0.73	0.99	0.040	0.86	0.73	1.00	0.048
A4: Compliance	0.93	0.81	1.07	0.329	0.93	0.81	1.07	0.293
A5: Modesty	0.83	0.71	0.97	0.021	0.82	0.70	0.96	0.016
A6: Tender-mindedness	0.89	0.77	1.03	0.117	0.88	0.76	1.03	0.103
C1: Competence	0.93	0.79	1.09	0.376	0.94	0.80	1.11	0.457
C2: Order	0.94	0.81	1.09	0.406	0.94	0.81	1.10	0.435
C3: Dutifulness	0.78	0.67	0.92	0.002	0.79	0.68	0.92	0.003
C4: Achievement striving	0.83	0.72	0.97	0.020	0.84	0.72	0.98	0.027
C5: Self-discipline	0.85	0.73	0.99	0.039	0.87	0.74	1.01	0.067
C6: Deliberation	0.97	0.83	1.14	0.740	0.97	0.83	1.14	0.709

Total N = 1,668, impaired N = 187. Model 1 includes age, sex, education, and form (form = personality self-reported versus interview) as covariates. Model 2 includes Model 1 covariates and obesity, smoking, hypertension, diabetes, depression, and APOE ɛ4 carrier status. HR, hazard ratios; CI, confidence interval. Bold p < 0.05.

Facets

Among the facets of neuroticism, depression had the strongest association, followed by impulsiveness and self-consciousness. Among the facets of conscientiousness, dutifulness had the strongest association, followed by achievement striving and self-discipline. Among other notable findings, ideas was the only facet of openness and warmth the only facet of extraversion that were associated with risk of impairment. For most facets, there were small differences between models 1 and 2, which suggests that the associations between personality facets and risk of impairment were mostly independent of APOE ɛ4, clinical, and behavioral risk factors. As a test of the generalizability of the associations between personality facets and risk of cognitive impairment across samples, we found an r = 0.67 (double-entry intraclass correlation, r = 0.64) between the HRs obtained for the 30 facets in the SardiNIA and the BLSA samples [25]. By comparison, the correlation between the HRs for the five domains in the SardiNIA and BLSA samples was r = 0.93 (double-entry intraclass correlation, r = 0.86). Supplementary Table 2 presents the coefficients from the SardiNIA and BLSA samples side-by-side.

Moderators

The interactions between each of the five major personality traits and sex, age, and APOE ɛ4 carrier status were tested to examine whether the associations varied by demographic and genetic risk groups. With one exception, all interactions were not significant, indicating that the associations were similar among men and women, across age, and those with and without the APOE ɛ4 risk variant. The one significant interaction was between conscientiousness and age (B = –0.196, SE = 0.087, p = 0.02).

DISCUSSION

In a rural sample followed for up to 18 years, this study found that higher neuroticism was a risk factor while higher conscientiousness, openness, and agreeableness were protective against the risk of cognitive impairment. At the facet level, the depression facet of neuroticism and the dutifulness facet of conscientiousness were among the strongest predictors. The effect of openness was mainly driven by the ideas facet (because it was the only facet of openness that was significantly associated with risk of impairment). Facets of agreeableness and the warmth facet of extraversion were also associated with reduced risk. The associations were robust when accounting for clinical and behavioral risk factors, including the APOE ɛ4 risk variant. The associations were also consistent across demographic groups, with no systematic evidence of moderation across sex, age, or APOE status. Given that the results were similar to other studies [1], it suggests that the association between personality and dementia risk does not depend on demographic factors or the APOE risk status.

The findings from this study suggest that the associations between personality and cognitive impairment are likely to generalize across diverse groups and populations. Indeed, the results from this rural sample with relatively low education were consistent with previous research on personality and dementia based on samples that generally had higher education and were from metropolitan areas. The effect sizes for neuroticism and conscientiousness were slightly weaker compared to the meta-analytic estimates [1]. However, we found that higher openness and agreeableness were protective, with effects slightly larger than those reported in the meta-analysis [1]. Of interest, the same meta-analysis observed that the associations were stronger for extraversion and weaker for agreeableness among European samples, but in this European sample, we found no association for extraversion and a stronger effect for agreeableness.

We did not find clear evidence that facets have larger predictive power compared to the respective broader domains. For example, we found effects for the facets of neuroticism that were similar to those of the overall domain, with only the depression facet (HR = 1.22) having an effect slightly stronger than the overall neuroticism domain (HR = 1.21). A similar pattern was observed for conscientiousness and its facets. Interestingly, dutifulness (a measure of responsibility, respect of duty and obligation, and dependability toward others) seems to be the only facet of conscientiousness facet consistently linked to risk of impairment across studies [25, 27]. Overall, multiple facets contribute to the neuroticism and conscientiousness associations with cognitive health, and there is not a single facet of these broad domains that fully explains the associations. A different pattern emerged for openness: Ideas, a measure of intellectual curiosity, was the facet driving the overall protective effect of openness, a finding that replicates the pattern observed in a previous study [25]. The warmth facet of extraversion had a protective effect, but it was not significant in a previous study [25]. Warmth is likely to help build social relations, which can protect late life cognition [17]. There were also some surprising null associations, such as anxiety (facet of neuroticism) and self-discipline (facet of conscientiousness) that were related to dementia risk in previous studies [25 –27]. While the specific facets that reach significance may vary across studies, an important finding was that the effect sizes of the facets seem to replicate across samples. Indeed, the HRs across the 30 facets obtained in the SardiNIA sample and a previous study [25] were highly correlated (r = 0.67). While the facets findings seem to generalize well across samples, it is also interesting to note that the pattern of replicability was more robust at the level of the five broad domains.

Multiple non-mutually exclusive mechanisms have been considered for the associations between personality and dementia risk [37, 38]. Personality may (a) be a risk/protective (etiological) factor and modulate resistance to neuropathology [39, 40]; (b) have a pathoplastic effect by modulating the timing and expression of the symptoms in the presence of neuropathology (resilience) [41, 42]; (c) share a genetic or environmental cause (common cause) [43, 44]; (d) be a prodromal feature or a response to emerging cognitive impairment (reverse causality) [45]. The observed associations could also be the result of residual or unmeasured confounding. An etiological role of personality could be indirect through more proximal risk factors. Indeed, personality traits are linked to health behaviors (e.g., smoking) [46, 47], lifestyles (e.g., cognitive and social engagement) [48, 49], and health conditions (e.g., depression) [50], which in turn are implicated in dementia [16, 17]. However, we found that accounting for some behavioral and clinical risk factors had little impact on the associations, suggesting that other pathways (e.g., inflammation, coping, loneliness) could playa role.

The results of this study could be further interpreted within the theoretical framework of cognitive reserve, defined as the individual differences in cognitive processes that allow some people to cope better than others with brain pathology [51]. Variables like education or IQ are often used as proxies of cognitive reserve, but personality traits can also modulate a person’s ability to cope with neurodegeneration [52]. For example, individuals high in openness tend to engage in novel and stimulating activities that are likely to build reserve over their lifespan [48, 53]. Individuals high in conscientiousness tend to be planful, disciplined, methodical, organized, and driven. Such traits can help create routines and provide structure in daily life, which over time can help sustain functioning and retain independence even in the face of physiological declines [54]. In people who score low on neuroticism, their better coping skills, lower stress reactivity, and emotional resilience can buffer the deficits associated with neuronal losses [55]. As noted above, personality traits are likely to build reserve also through health behaviors [46, 47] and social connectedness [56]. This idea that personality modulates reserve has also gained support from neuropathology studies: in the presence of neuropathology, a resilient personality was found to help retain cognitive function and delay the onset of cognitive impairment [41, 42].

Strengths and limitations

The study had several strengths, including the focus on a rural sample, the inclusion of participants with a broad range of education levels, the use of an in-depth measure of personality that includes facets as well as domains, and a mean follow-up of over ten years. And while the MMSE is a reliable, valid, and widely used cognitive screening instrument in research and clinical practice, a limitation of this study was the lack of clinical diagnosis of dementia. The MMSE likely misclassified some participants, and even when correctly classified as cognitively impaired, such impairment could be due to causes other than dementia, such as stroke or head trauma. The lack of clinical diagnosis is likely to increase measurement error or other biases if differentially related to the cognitive and personality scores. However, it is unlikely that the MMSE introduced major systematic biases, given that the results from this study are consistent with those that used other methods, such as clinical diagnosis [25, 26] or health records [12]. While different MMSE cutoffs have been used in previous studies, it should be reassuring that the associations were similar when using the continuous MMSE score in this sample and similar to the meta-analytic findings [1]. Other limitations were the relatively infrequent MMSE assessment (every 3–4 years) and the lack of cognitive assessment at baseline. We cannot exclude that the associations are driven partly by reverse causality (i.e., Alzheimer’s disease neuropathology could cause changes in personality before the onset of other clinical signs), and there are changes in personality with the onset and across the clinical stages of Alzheimer’s disease and related dementias [57 –59]. However, contrary to the reverse causality hypothesis, a study with over 30 years of follow-up found no increases in neuroticism or declines in conscientiousness during the preclinical stage of dementia [45]. Furthermore, a meta-analysis found no relation between the length of follow-up and the strength of the associations [1]. Moreover, even when assessed in adolescents, personality is prospectively related to cognitive functioning and dementia risk in adulthood[60, 61].

Conclusions

By examining the prospective association of personality with repeated measures of global cognition over up to 18 years of follow-up, this study advances knowledge on personality and dementia for at least two reasons. First, the findings support the generalizability of the associations between personality and cognitive health across diverse communities. It adds to evidence from Colombia [62] and Brazil [63] that the associations replicate across diverse communities. Second, it provides a more granular understanding of the associations between personality and cognitive impairment by examining facets. Facets of personality produce replicable results, and while multiple facets likely drive the associations of neuroticism and conscientiousness, the ideas facet seems responsible for the protective effect of openness. The effects sizes of personality are similar to other risk factors [25], such as physical activity, emphasizing the importance of considering individual differences in personality to understand the etiology of neurodegenerative diseases and developing interventions aimed at dementia prevention. The replicable evidence supports the need for more research to test the role of personality in the etiology of dementia, risk reduction, and tailored treatments [64 –66].

Footnotes

ACKNOWLEDGMENTS

The SardiNIA study is supported by the Intramural Program of the National Institute on Aging, National Institutes of Health. This work has been supported by the National Institute on Aging of the National Institutes of Health under Award Numbers R01AG068093 and R01AG053297 and 75N95021C00012. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Aschwanden

, Strickhouser

, Luchetti

, Stephan

, Sutin

, Terracciano

(2021) Is personality associated with dementia risk? A meta-analytic investigation. Ageing Res Rev 67, 101269.

Costa

Jr , McCrae

(1992) Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual, Psychological Assessment Resources, Odessa, FL.

Schroots

(1996) Theoretical developments in the psychology of aging. Gerontologist 36, 742–748.

Mroczek

, Spiro

III A

, Griffin

(2006) Personality and aging. In Handbook of the Psychology of Aging, Elsevier, pp. 363–377.

Roberts

, Kuncel

, Shiner

, Caspi

, Goldberg

(2007) The power of personality: The comparative validity of personality traits, socioeconomic status, and cognitive ability for predicting important life outcomes. Perspect Psychol Sci 2, 313–345.

Terracciano

, Aschwanden

, Stephan

, Cerasa

, Passamonti

, Toschi

, Sutin

(2021) Neuroticism and risk of Parkinson’s disease: A meta-analysis. Mov Disord 36, 1863–1870.

Duchek

, Balota

, Storandt

, Larsen

(2007) The power of personality in discriminating between healthy aging and early-stage Alzheimer’s disease. J Gerontol B Psychol Sci Soc Sci 62, P353–P361.

Duberstein

, Chapman

, Tindle

, Sink

, Bamonti

, Robbins

, Jerant

, Franks

(2011) Personality and risk for Alzheimer’s disease in adults 72 years of age and older: A 6-year follow-up. Psychol Aging 26, 351–362.

Johansson

, Guo

, Duberstein

, Hallstrom

, Waern

, Ostling

, Skoog

(2014) Midlife personality and risk of Alzheimer disease and distress: A 38-year follow-up. Neurology 83, 1538–1544.

10.

Kaup

, Harmell

, Yaffe

(2019) Conscientiousness is associated with lower risk of dementia among black and white older adults. Neuroepidemiology 52, 86–92.

11.

Wilson

, Barnes

, Bennett

, Li

, Bienias

, Mendes de Leon

, Evans

(2005) Proneness to psychological distress and risk of Alzheimer disease in a biracial community. Neurology 64, 380–382.

12.

Terracciano

, Aschwanden

, Passamonti

, Toschi

, Stephan

, Luchetti

, Lee

, Sesker

, O’Súilleabháin

, Sutin

(2021) Is neuroticism differentially associated with risk of Alzheimer’s disease, vascular dementia, and frontotemporal dementia? . J Psychiatr Res 138, 34–40.

13.

Anstey

, Peters

, Zheng

, Barnes

, Brayne

, Brodaty

, Chalmers

, Clare

, Dixon

, Dodge

(2020) Future directions for dementia risk reduction and prevention research: An international research network on dementia prevention consensus. J Alzheimers Dis 78, 3–12.

14.

Rauthmann

(2020) Personality science: United in diversity. Personal Sci 1, 1–25.

15.

Russ

, Batty

, Hearnshaw

, Fenton

, Starr

(2012) Geographical variation in dementia: Systematic review with meta-analysis. Int J Epidemiol 41, 1012–1032.

16.

, Xu

, Tan

, Andrieu

, Suckling

, Evangelou

, Pan

, Zhang

, Jia

, Feng

(2020) Evidence-based prevention of Alzheimer’s disease: Systematic review and meta-analysis of 243 observational prospective studies and 153 randomised controlled trials. J Neurol Neurosurg Psychiatry 91, 1201–1209.

17.

World Health Organization (2019) Risk reduction of cognitive decline and dementia: WHO guidelines.

18.

Pilia

, Chen

, Scuteri

, Orrú

, Albai

, Dei

, Lai

, Usala

, Lai

, Loi

, Mameli

, Vacca

, Deiana

, Masala

, Cao

, Najjar

, Terracciano

, Nedorezov

, Sharov

, Zonderman

, Abecasis

, Costa

, Lakatta

, Schlessinger

(2006) Heritability of cardiovascular and personality traits in 6,148 Sardinians. PloS Genet 2, e132.

19.

Piscopo

, Manfredi

, Malvezzi-Campeggi

, Crestini

, Spadoni

, Cherchi

, Deiana

, Piras

, Confaloni

(2006) Genetic study of Sardinian patients with Alzheimer’s disease. Neurosci Lett 398, 124–128.

20.

Singh

, Singh

, Mastana

(2006) APOE distribution in world populations with new data from India and the UK. Ann Hum Biol 33, 279–308.

21.

Farrer

, Cupples

, Haines

, Hyman

, Kukull

, Mayeux

, Myers

, Pericak-Vance

, Risch

, Van Duijn

(1997) Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease: A meta-analysis. JAMA 278, 1349–1356.

22.

Buettner

(2012) The Blue Zones: 9 lessons for living longer from the people who’ve lived the longest, National Geographic Books.

23.

Stewart

, Mõttus

, Seeboth

, Soto

, Johnson

(2021) The finer details? The predictability of life outcomes from Big Five domains, facets, and nuances. J Pers 90, 167–182.

24.

Sutin

, Terracciano

, Kitner-Triolo

, Uda

, Schlessinger

, Zonderman

(2011) Personality traits prospectively predict verbal fluency in a lifespan sample. Psychol Aging 26, 994–999.

25.

Terracciano

, Sutin

, An

, O’Brien

, Ferrucci

, Zonderman

, Resnick

(2014) Personality and risk of Alzheimer’s disease: New data and meta-analysis. Alzheimers Dement 10, 179–186.

26.

Wilson

, Begeny

, Boyle

, Schneider

, Bennett

(2011) Vulnerability to stress, anxiety, and development of dementia in old age. Am J Geriatr Psychiatry 19, 327–334.

27.

Sutin

, Stephan

, Terracciano

(2018) Facets of conscientiousness and risk of dementia. Psychol Med 48, 974–982.

28.

Kekäläinen

, Terracciano

, Sipilä

, Kokko

(2020) Personality traits and physical functioning: A cross-sectional multimethod facet-level analysis. Eur Rev Aging Phys Act 17, 20.

29.

Costa

Jr , Terracciano

, Uda

, Vacca

, Mameli

, Pilia

, Zonderman

, Lakatta

, Schlessinger

, McCrae

(2007) Personality traits in Sardinia: Testing founder population effects on trait means and variances. Behav Genet 37, 376–387.

30.

Terracciano

(2003) The Italian version of the NEO PI-R: Conceptual and empirical support for the use of targeted rotation. Pers Indivd Dif 35, 1859–1872.

31.

McCrae

, Terracciano

, Personality Profiles of Cultures Project (2005) Universal features of personality traits from the observer’s perspective: Data from 50 cultures. J Pers Soc Psychol 88, 547–561.

32.

Folstein

, Folstein

, McHugh

(1975) Mini-Mental State. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

33.

Kochhann

, Varela

, Lisboa

CSM

, Chaves

MLF

(2010) The Mini Mental State Examination: Review of cutoff points adjusted for schooling in a large Southern Brazilian sample. Dement Neuropsychol 4, 35–41.

34.

Mazzoni

, Ferroni

, Lombardi

, Del Torto

, Vista

, Moretti

(1992) Mini-Mental State Examination (MMSE): Sensitivity in an Italian sample of patients with dementia. Ital J Neurol Sci 13, 323–329.

35.

Perneger

(1998) What’s wrong with Bonferroni adjustments. BMJ 316, 1236–1238.

36.

Lykken

(1968) Statistical significance in psychological research. Psychol Bull 70, 151–159.

37.

Terracciano

, Sutin

(2019) Personality and Alzheimer’s disease: An integrative review. Personal Disord 10, 4–12.

38.

Segerstrom

(2020) Personality and incident Alzheimer’s disease: Theory, evidence, and future directions. J Gerontol B Psychol Sci Soc Sci 75, 513–521.

39.

Terracciano

, Bilgel

, Aschwanden

, Luchetti

, Stephan

, Moghekar

, Wong

, Ferrucci

, Sutin

, Resnick

(2022) Personality associations with amyloid and tau: Results from the Baltimore Longitudinal Study of Aging and meta-analysis. Biol Psychiatry 91, 359–369.

40.

Duchek

, Aschenbrenner

, Fagan

, Benzinger

, Morris

, Balota

(2020) The relation between personality and biomarkers in sensitivity and conversion to Alzheimer-type dementia. J Int Neuropsychol Soc 26, 596–606.

41.

Graham

, James

, Jackson

, Willroth

, Boyle

, Wilson

, Bennett

, Mroczek

(2021) Associations between personality traits and cognitive resilience in older adults. J Gerontol B Psychol Sci Soc Sci 76, 6–19.

42.

Terracciano

, Iacono

, O’Brien

, Troncoso

, An

, Sutin

, Ferrucci

, Zonderman

, Resnick

(2013) Personality and resilience to Alzheimer’s disease neuropathology: A prospective autopsy study. Neurobiol Aging 34, 1045–1050.

43.

Stephan

, Sutin

, Luchetti

, Caille

, Terracciano

(2018) Polygenic score for Alzheimer disease and cognition: The mediating role of personality. J Psychiatr Res 107, 110–113.

44.

Okbay

, Baselmans

, De Neve

, Turley

, Nivard

, Fontana

, Meddens

, Linner

, Rietveld

, Derringer

, et al. (2016) Genetic variants associated with subjective well-being, depressive symptoms, and neuroticism identified through genome-wide analyses. Nat Genet 48, 624–633.

45.

Terracciano

, An

, Sutin

, Thambisetty

, Resnick

(2017) Personality change in the preclinical phase of Alzheimer disease. JAMA Psychiatry 74, 1259–1265.

46.

Hakulinen

, Hintsanen

, Munafò

, Virtanen

, Kivimäki

, Batty

, Jokela

(2015) Personality and smoking: Individual-participant meta-analysis of nine cohort studies. Addiction 110, 1844–1852.

47.

Sutin

, Stephan

, Luchetti

, Artese

, Oshio

, Terracciano

(2016) The five-factor model of personality and physical inactivity: A meta-analysis of 16 samples. J Res Pers 63, 22–28.

48.

Stephan

, Boiche

, Canada

, Terracciano

(2014) Association of personality with physical, social, and mental activities across the lifespan: Findings from US and French samples. Br J Psychol 105, 564–580.

49.

Rohrer

, Lucas

, Donnellan

, Schlegel

(2018) Only so many hours: Correlations between personality and daily time use in a representative German panel. Collabra Psychol 4, 1.

50.

Klein

, Kotov

, Bufferd

(2011) Personality and depression: Explanatory models and review of the evidence. Annu Rev Clin Psychol 7, 269–295.

51.

Stern

(2009) Cognitive reserve. Neuropsychologia 47, 2015–2028.

52.

Colombo

, Piromalli

, Pins

, Taylor

, Fabio

(2020) The relationship between cognitive reserve and personality traits: A pilot study on a healthy aging Italian sample. Aging Clin Exp Res 32, 2031–2040.

53.

Franchow

, Suchy

, Thorgusen

, Williams

(2013) More than education: Openness to experience contributes to cognitive reserve in older adulthood. Aging Sci 1, 109.

54.

Stephan

, Fundenberger

, Sutin

, Terracciano

(2021) Cross-sectional and prospective association between personality traits and IADL/ADL limitations. Psychol Aging 36, 309–321.

55.

Connor-Smith

, Flachsbart

(2007) Relations between personality and coping: A meta-analysis. J Pers Soc Psychol 93, 1080–1107.

56.

Buecker

, Maes

, Denissen

, Luhmann

(2020) Loneliness and the Big Five personality traits: A meta–analysis. Eur J Pers 34, 8–28.

57.

Islam

, Mazumder

, Schwabe-Warf

, Stephan

, Sutin

, Terracciano

(2019) Personality changes with dementia from the informant perspective: New data and meta-analysis. J Am Med Dir Assoc 20, 131–137.

58.

Torrente

, Pose

, Gleichgerrcht

, Torralva

, Lopez

, Cetkovich-Bakmas

, Manes

(2014) Personality changes in dementia: Are they disease specific and universal? . Alzheimer Dis Assoc Disord 28, 261–268.

59.

Caselli

, Langlais

, Dueck

, Henslin

, Johnson

, Woodruff

, Hoffman-Snyder

, Locke

DEC

(2018) Personality changes during the transition from cognitive health to mild cognitive impairment. J Am Geriatr Soc 66, 671–678.

60.

Chapman

, Huang

, Peters

, Horner

, Manly

, Bennett

, Lapham

(2020) Association between high school personality phenotype and dementia 54 years later in results from a national US sample. JAMA Psychiatry 77, 148–154.

61.

Sutin

, Stephan

, Luchetti

, Aschwanden

, Sesker

, O’Súilleabháin

, Terracciano

TerraccianoA(2021) Self-reported and mother-rated personality traits at age 16 are associated with cognitive function measured concurrently and 30 years later. Psychol Med, doi: 10.1017/S0033291721000672.

62.

Baena

, Bocanegra

, Torres

, Vila-Castelar

, Guzmán-Vélez

, Fox-Fuller

, Gatchel

, Sánchez

, Pluim

, Ramirez-Gómez

, Martínez

, Pineda

, Lopera

, Quiroz

(2021) Neuroticism is associated with tau pathology in cognitively unimpaired individuals with autosomal dominant Alzheimer’s disease. J Alzheimers Dis 82, 1809–1822.

63.

Capuano

, Wilson

, Leurgans

, Sampaio

, Barnes

, Farfel

, Bennett

(2021) Neuroticism, negative life events, and dementia in older White and Black Brazilians. Int J Geriatr Psychiatry 36, 901–908.

64.

Rouch

, Dorey

, Padovan

, Trombert-Paviot

, Benoit

, Laurent

, Boublay

, Krolak-Salmon

(2019) Does personality predict behavioral and psychological symptoms of dementia? Results from PACO Prospective Study. J Alzheimers Dis 69, 1099–1108.

65.

Rouch

, Pongan

, Leveque

, Tillmann

, Trombert

, Getenet

, Auguste

, Krolak-Salmon

, Laurent

, Dorey

(2018) Personality modulates the efficacy of art intervention on chronic pain in a population of patients with Alzheimer’s disease. {J Alzheimers Dis 63, 617–624.

66.

Khayoun

, Devick

, Chandler

, Shandera-Ochsner

, De Wit

, Cuc

, Smith

, Locke

(2021) The impact of patient and partner personality traits on learning success for a cognitive rehabilitation intervention for patients with MCI. Neuropsychol Rehabil. doi: 10.1080/09602011.2021.1948872.