Personality Predictors of Dementia Diagnosis and Neuropathic Burden: A Mega-Analysis
University of California, DavisFeinberg School of Medicine
Feinberg School of Medicine, Northwestern University
Feinberg School of Medicine
27 July 2023
Abstract
“INTRODUCTION: The extent to which the Big Five personality traits and subjective well-being (SWB) are discriminatory predictors of clinical manifestation of dementia versus dementia-related neuropathology is unclear.”
“METHODS: Using data from eight independent studies (Ntotal=44,531; baseline Mage= years, % female), Bayesian multilevel models tested whether traits and SWB differentially predicted neuropsychological and neuropathological characteristics of dementia.”
“RESULTS: Adjusting for sociodemographic and health covariates, synthesized and individual study results indicate that high neuroticism, low conscientiousness, and high negative affect were associated with increased risk of long-term dementia diagnosis.”
“DISCUSSION: This multi-study project provides robust, conceptually replicated evidence that psychosocial factors are strong predictors of dementia diagnosis, but differentially associated with neuropathology at autopsy as a function of premorbid dementia diagnoses in some samples. These results suggest the possible importance of brain maintenance or test performance, while also highlighting the need for ongoing data collection efforts to disentangle these complex relationships.”
“METHODS: Using data from eight independent studies (Ntotal=44,531; baseline Mage= years, % female), Bayesian multilevel models tested whether traits and SWB differentially predicted neuropsychological and neuropathological characteristics of dementia.”
“RESULTS: Adjusting for sociodemographic and health covariates, synthesized and individual study results indicate that high neuroticism, low conscientiousness, and high negative affect were associated with increased risk of long-term dementia diagnosis.”
“DISCUSSION: This multi-study project provides robust, conceptually replicated evidence that psychosocial factors are strong predictors of dementia diagnosis, but differentially associated with neuropathology at autopsy as a function of premorbid dementia diagnoses in some samples. These results suggest the possible importance of brain maintenance or test performance, while also highlighting the need for ongoing data collection efforts to disentangle these complex relationships.”
Chapter 1 Workspace
In this section, we’ll set up everything we need to clean data in the next section. This includes:
- Loading in all packages
- Loading in the codebook
- Setting up data frames for personality traits / well-being, outcomes, covariates, and moderators, so that we can more easily rename their short-hand names to production ready ones later
- Loading in and rendering html tables of some descriptives, measures, etc.
1.1 Packages
First, let’s load in the packages. Note the descriptions for each commented next to them.
library(knitr) # knit documents
library(kableExtra) # formatted tables
library(readxl) # read excel files
library(haven) # read spss files
library(broom.mixed) # summaries of models
library(rstan) # bayes underpinnings
library(tidybayes) # pretty bayes draws and plots
library(cowplot) # piece plots together
library(plyr) # data wrangling
library(tidyverse) # data wrangling
library(brms) # bayesian models
library(furrr) # parallel purrr mapping
library(psych) # psychometrics1.2 Directory Path
We have three different directories:
1. data_path stores the raw that that cannot be shared per data use agreements
2. res_path includes the GitHub link where shareable objects can be found
3. local_path is mostly used to save files as they render, or in some limited cases of objects that hold raw data, to access those objects that can’t be shared
1.3 Codebook
Each study has a separate codebook indexing covariate, moderator, personality, and outcome variables. Moreover, these codebooks contain information about the original scale of the variable, any recoding of the variable (including binarizing outcomes, changing the scale, and removing missing data), reverse coding of scale variables, categories, etc.
# list of all codebook sheets
url <- "https://github.com/emoriebeck/personality-dementia-neuropath/raw/master/codebooks/master_codebook_09.04.20.xlsx?raw=true"
destfile <- "master_codebook_09.04.20.xlsx"
# destfile <- sprintf("%s/codebooks/%s", local_path, destfile)
curl::curl_download(url, destfile)
sheets <- excel_sheets(destfile)
# function for reading in sheets
read_fun <- function(x){
read_xlsx(destfile, sheet = x)
}
# read in sheets and index source
codebook <- tibble(
study = sheets,
codebook = map(study, read_fun)
)
## short and long versions of names of all categories for later use
studies <- c("ROS", "RADC-MAP", "EAS", "ADRC" , "SATSA", "HRS", "LISS", "GSOEP")
studies_long <- c("ROS", "Rush-MAP", "EAS", "WUSM-MAP", "SATSA", "HRS", "LISS", "GSOEP")
stdcolors <- tibble(
studies = c("Overall", studies)
, studies_long = c("Overall", studies_long)
, std_text = str_remove_all(studies, "[-]")
, colors = c("black", "#332288", "#88ccee", "#44aa99", "#117733", "#999933", #"#ddcc77", "#cc6677",
"#332288", "#88ccee", "#44aa99")#, "#117733", "#999933", "#ddcc77")
, lt = c(rep("solid", 6), rep("dotted", 3)))
traits <- codebook$codebook[[2]] %>% filter(category == "pers") %>%
select(long_name = Construct, short_name = name); traits## # A tibble: 8 × 2
## long_name short_name
## <chr> <chr>
## 1 Extraversion E
## 2 Agreeableness A
## 3 Conscientiousness C
## 4 Neuroticism N
## 5 Openness to Experience O
## 6 Positive Affect PA
## 7 Negative Affect NA
## 8 Satisfaction with Life SWLoutcomes <- codebook$codebook[[2]] %>% filter(category == "out") %>%
select(long_name = Construct, short_name = name, link, colnm); outcomes## # A tibble: 11 × 4
## long_name short_name link colnm
## <chr> <chr> <chr> <chr>
## 1 Incident Dementia Diagnosis dementia factor OR [CI]
## 2 Braak Stage braak continuous b [CI]
## 3 CERAD cerad continuous b [CI]
## 4 Lewy Body Disease lewyBodyDis factor OR [CI]
## 5 Gross Cerebral Infarcts vsclrInfrcts factor OR [CI]
## 6 Gross Cerebral Microinfarcts vsclrMcrInfrcts factor OR [CI]
## 7 Cerebral Atherosclerosis atherosclerosis continuous b [CI]
## 8 Cerebral Amyloid Angiopathy angiopathy continuous b [CI]
## 9 Arteriolosclerosis arteriolosclerosis continuous b [CI]
## 10 Hippocampal Sclerosis hipSclerosis factor OR [CI]
## 11 TDP-43 tdp43 factor OR [CI]moders <- codebook$codebook[[2]] %>% filter(category == "mod") %>%
select(long_name = Construct, short_name = name, short_term = old_term, long_term = new_term); moders## # A tibble: 6 × 4
## long_name short_name short_term long_term
## <chr> <chr> <chr> <chr>
## 1 None none p_value Personality
## 2 Age age age Age
## 3 Gender gender gender1 Gender (Male v Female)
## 4 Education education education Education (Years)
## 5 Cognition cognition cognition Cognition
## 6 Dementia Diagnosis dementia dementia Dementia Diagnosis (No v Yes)covars <- codebook$codebook[[2]] %>% filter(category == "covariates") %>%
select(long_name = Construct, short_name = name, desc = new_term); covars## # A tibble: 8 × 3
## long_name short_name desc
## <chr> <chr> <chr>
## 1 Unadjusted unadjusted Unadjusted indicates no covariates were…
## 2 Fully Adjusted fully Fully adjusted models include age, gend…
## 3 Shared Covariates Adjusted shared Shared covariates adjusted models Inclu…
## 4 Standard Covariates Adjusted standard Standard covariates adjusted models inc…
## 5 All But One Covariate Adjusted butOne All but one covariate adjusted models i…
## 6 Shared Covariates Adjusted (With Dementia Diagnosis) shareddx Shared covariates with dementia adjuste…
## 7 Standard Covariates Adjusted (With Dementia Diagnosis) standarddx Standard covariates with dementia adjus…
## 8 Shared Covariates Adjusted (With Prediction Interval) sharedint Shared covariates with prediction inter…# used personality waves
url <- "https://github.com/emoriebeck/personality-dementia-neuropath/raw/master/codebooks/tables.xlsx?raw=true"
destfile <- "tables.xlsx"
# destfile <- sprintf("%s/codebooks/%s", local_path, destfile)
curl::curl_download(url, destfile)
p_waves <- read_xlsx(destfile, sheet = "Table 2")1.4 Tables
1.4.1 Table S1
Below, I create Table S1, which includes information the personality and well-being scales used in each study:
p_tab <- p_waves %>%
select(Study, everything(), -p_item, -Used) %>%
filter(Study != "BLSA") %>%
mutate(Measure = factor(Measure, traits$long_name)) %>%
arrange(Study, Measure)
rs <- p_tab %>%
group_by(Study) %>%
tally() %>%
mutate(end = cumsum(n), start = lag(end) + 1, start = ifelse(is.na(start), 1, start))
p_tab <- p_tab %>%
select(-Study) %>%
kable(.
, "html"
, caption = "<strong>Table S1</strong><br><em>Personality Trait and Subjective Well-Being Measurement Inventories, Scales, and Assessments Across Samples</em>"
, escape = F
, col.names = paste0("<strong>", colnames(p_tab)[-1], "</strong>")
, align = c("r", "l", "l", "c")
) %>%
kable_classic(full_width = F, html_font = "Times New Roman")
for (i in 1:nrow(rs)){
p_tab <- p_tab %>% kableExtra::group_rows(rs$Study[i], rs$start[i], rs$end[i])
}
p_tab| Measure | Source | Scale | Used (Available) |
|---|---|---|---|
| EAS | |||
| Extraversion | 24 items from the IPIP NEO | 1 “strongly disagree” to 5 “strongly agree” | 2004 (Annual Follow-ups) |
| Agreeableness | 24 items from the IPIP NEO | 1 “strongly disagree” to 5 “strongly agree” | 2004 (Annual Follow-ups) |
| Conscientiousness | 24 items from the IPIP NEO | 1 “strongly disagree” to 5 “strongly agree” | 2004 (Annual Follow-ups) |
| Neuroticism | 24 items from the IPIP NEO | 1 “strongly disagree” to 5 “strongly agree” | 2004 (Annual Follow-ups) |
| Openness to Experience | 24 items from the IPIP NEO | 1 “strongly disagree” to 5 “strongly agree” | 2004 (Annual Follow-ups) |
| Positive Affect | — | — | — |
| Negative Affect | — | — | — |
| Satisfaction with Life | — | — | — |
| GSOEP | |||
| Extraversion | 3 items from the 15 item BFI-S (John, Naumann, & Soto, 2008, and Lang, Lüdtke, & Asendorpf, 2001) | 1 “does not apply at all” to 7 ” applies perfectly” | 2005 (2005, 2009, 2013, 2017) |
| Agreeableness | 3 items from the 15 item BFI-S (John, Naumann, & Soto, 2008, and Lang, Lüdtke, & Asendorpf, 2001) | 1 “does not apply at all” to 7 ” applies perfectly” | 2005 (2005, 2009, 2013, 2017) |
| Conscientiousness | 3 items from the 15 item BFI-S (John, Naumann, & Soto, 2008, and Lang, Lüdtke, & Asendorpf, 2001) | 1 “does not apply at all” to 7 ” applies perfectly” | 2005 (2005, 2009, 2013, 2017) |
| Neuroticism | 3 items from the 15 item BFI-S (John, Naumann, & Soto, 2008, and Lang, Lüdtke, & Asendorpf, 2001) | 1 “does not apply at all” to 7 ” applies perfectly” | 2005 (2005, 2009, 2013, 2017) |
| Openness to Experience | 3 items from the 15 item BFI-S (John, Naumann, & Soto, 2008, and Lang, Lüdtke, & Asendorpf, 2001) | 1 “does not apply at all” to 7 ” applies perfectly” | 2005 (2005, 2009, 2013, 2017) |
| Positive Affect | 1 item (“Frequency of being happy in the last 4 weeks”) | 1 “very seldom” to 5 “very often” | 2007 (2007-2017) |
| Negative Affect | 3 items (angry, sad, worried) | 1 “very seldom” to 5 “very often” | 2007 (2007-2017) |
| Satisfaction with Life | SWLS (Diener, Emmons, Larsen, & Griffin, 1985) | 0 “low” to 10 “high” | 2005 (1984-2017) |
| HRS | |||
| Extraversion | 5 adjectives from a 25 adjective checklist (Lachman & Weaver, 1997) | **1 “a lot” to 4 “not at all” | 2006/8 (2006/8, 2010/12, 2014/16) |
| Agreeableness | 5 adjectives from a 25 adjective checklist (Lachman & Weaver, 1997) | **1 “a lot” to 4 “not at all” | 2006/8 (2006/8, 2010/12, 2014/16) |
| Conscientiousness | 5 adjectives from a 25 adjective checklist (Lachman & Weaver, 1997) | **1 “a lot” to 4 “not at all” | 2006/8 (2006/8, 2010/12, 2014/16) |
| Neuroticism | 5 adjectives from a 25 adjective checklist (Lachman & Weaver, 1997) | **1 “a lot” to 4 “not at all” | 2006/8 (2006/8, 2010/12, 2014/16) |
| Openness to Experience | 5 adjectives from a 25 adjective checklist (Lachman & Weaver, 1997) | **1 “a lot” to 4 “not at all” | 2006/8 (2006/8, 2010/12, 2014/16) |
| Positive Affect | PANAS-X (Watson & Clark, 1994) | *1 “very much” to 5 “not at all” | 2006-2016 |
| Negative Affect | PANAS-X (Watson & Clark, 1994) | *1 “very much” to 5 “not at all” | 2006-2016 |
| Satisfaction with Life | SWLS (Diener, Emmons, Larsen, & Griffin, 1985) | 1 “strongly disagree” to 7 “strongly agree” | 2006/8 (2006/8, 2010/12, 2014/16) |
| LISS | |||
| Extraversion | 10 items from the 50 item IPIP-50 (Goldberg, 1992) | 1 “very inaccurate” to 5 “very accurate” | 2008 (2008-2018) |
| Agreeableness | 10 items from the 50 item IPIP-50 (Goldberg, 1992) | 1 “very inaccurate” to 5 “very accurate” | 2008 (2008-2018) |
| Conscientiousness | 10 items from the 50 item IPIP-50 (Goldberg, 1992) | 1 “very inaccurate” to 5 “very accurate” | 2008 (2008-2018) |
| Neuroticism | 10 items from the 50 item IPIP-50 (Goldberg, 1992) | 1 “very inaccurate” to 5 “very accurate” | 2008 (2008-2018) |
| Openness to Experience | 10 items from the 50 item IPIP-50 (Goldberg, 1992) | 1 “very inaccurate” to 5 “very accurate” | 2008 (2008-2018) |
| Positive Affect | 10 items (e.g. “interested”) | 1 “not at all” to 7 “extremely” | 2008 (2008-2018) |
| Negative Affect | 10 items (e.g. “distressed”) | 1 “not at all” to 7 “extremely” | 2008 (2008-2018) |
| Satisfaction with Life | SWLS (Diener, Emmons, Larsen, & Griffin, 1985) | 1 “strongly disagree” to 7 “strongly agree” | 2008 (2008-2018) |
| RUSH-MAP | |||
| Extraversion | 6 items from the NEO Five Factor Inventory | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Agreeableness | — | — | — |
| Conscientiousness | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Neuroticism | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Openness to Experience | — | — | — |
| Positive Affect | 1 item “Overall, how happy are you?” | *1 “Very happy” to 4 “Not happy at all” | Baseline (Annual Clinical Followups) |
| Negative Affect | PANAS-X (Watson & Clark, 1994) | 1 “Very slightly or not at al” to 5 “Extremely” | Baseline (Annual Clinical Followups) |
| Satisfaction with Life | SWLS (Diener, Emmons, Larsen, & Griffin, 1985) | *1 “Strongly agree” to 7 “Strongly Disagree” | Baseline (Annual Clinical Followups) |
| RUSH-ROS | |||
| Extraversion | 6 items from the NEO Five Factor Inventory | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Agreeableness | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Conscientiousness | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Neuroticism | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Openness to Experience | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Baseline) |
| Positive Affect | 1 item “Overall, how happy are you?” | *1 “Very happy” to 4 “Not happy at all” | Baseline (Annual Clinical Followups) |
| Negative Affect | — | — | — |
| Satisfaction with Life | SWLS (Diener, Emmons, Larsen, & Griffin, 1985) | *1 “Strongly agree” to 7 “Strongly Disagree” | Baseline (Annual Clinical Followups) |
| SATSA | |||
| Extraversion | 9 items from the Eysenck Personality Inventory | **1 “exactly right” to 5 “not right at all” | 1984 (1984, 1987, 1989, 1990, 1992, 1993, 1999, 2002, 2004, 2005, 2007) |
| Agreeableness | 10 items | **1 “exactly right” to 5 “not right at all” | 1984 (1984) |
| Conscientiousness | 10 items | **1 “exactly right” to 5 “not right at all” | 1984 (1984) |
| Neuroticism | 9 items from the Eysenck Personality Inventory | **1 “exactly right” to 5 “not right at all” | 1984 (1984, 1987, 1989, 1990, 1992, 1993, 1999, 2002, 2004, 2005, 2007) |
| Openness to Experience | 25 items from the NEO Personality Inventory | **1 “exactly right” to 5 “not right at all” | 1984 (1984) |
| Positive Affect | 5 items (e.g., “calm”, “harmonious”) | **1 “exactly right” to 5 “not right at all” | 1984 (1984, 1987, 1989, 1990, 1992, 1993, 1995) |
| Positive Affect | 6 items (e.g., “worried”, “tense”) | **1 “exactly right” to 5 “not right at all” | 1984 (1984, 1987, 1989, 1990, 1992, 1993, 1995) |
| Satisfaction with Life | 13 items | **1 “exactly right” to 5 “not right at all” | 1984 (1984, 1987, 1989, 1990, 1993, 2004, 2007) |
| WUSM-MAP | |||
| Extraversion | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Clinical Follow-ups) |
| Agreeableness | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Clinical Follow-ups) |
| Conscientiousness | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Clinical Follow-ups) |
| Neuroticism | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Clinical Follow-ups) |
| Openness to Experience | 12 items from the NEO Five-Factor Inventory (NEO-FFI; Costa & McCrae, 1992) | 1 “strongly disagree” to 5 “strongly agree” | Baseline (Clinical Follow-ups) |
| Positive Affect | — | — | — |
| Negative Affect | — | — | — |
| Satisfaction with Life | — | — | — |
1.4.2 Table S2
Next, I create table S2, which indicates which measures were in each study, including which cognitive measures were in each sample.
meas <- read_xlsx(destfile, sheet = "Table 1") %>%
select(-BLSA)
rs <- meas %>%
group_by(Category) %>%
tally() %>%
mutate(end = cumsum(n), start = lag(end) + 1, start = ifelse(is.na(start), 1, start))
meas <- meas %>%
select(-Category) %>%
kable(.
, "html"
, caption = "<strong>Table S2</strong><br><em>List of Measures Across Samples</em>"
, escape = F
, col.names = paste0("<strong>", colnames(meas)[-1], "</strong>")
, align = c("l", rep("c", 8))
) %>%
kable_classic(full_width = F, html_font = "Times New Roman")
for (i in 1:nrow(rs)){
meas <- meas %>% kableExtra::group_rows(rs$Category[i], rs$start[i], rs$end[i])
}
meas| Measure | GSOEP | HRS | LISS | SATSA | RUSH-MAP | RUSH-ROS | ADRC-MAP | EAS |
|---|---|---|---|---|---|---|---|---|
| Cognitive | ||||||||
| Extraversion | X | X | X | X | X | X | X | X |
| Agreeableness | X | X | X | X | X | X | ||
| Conscientiousness | X | X | X | X | X | X | X | |
| Neuroticism | X | X | X | X | X | X | X | X |
| Openness to Experience | X | X | X | X | X | X | X | |
| Satisfaction with Life | X | X | X | X | X | X | X | |
| Positive Affect | X | X | X | X | X | |||
| Negative Affect | X | X | X | X | ||||
| Self-Reported Dementia | X | X | X | X | X | X | X | X |
| Braak Stage | X | X | X | X | ||||
| CERAD | X | X | X | |||||
| Lewy Body Disease | X | X | X | X | ||||
| Gross Cerebral Infarcts | X | X | X | |||||
| Gross Cerebral Microinfarcts | X | X | X | |||||
| Cerebral Atherosclerosis | X | X | X | |||||
| Cerebral Amyloid Angiopathy | X | X | X | |||||
| Covariates | ||||||||
| Arteriolosclerosis | X | X | X | |||||
| Hippocampal Sclerosis | X | X | X | X | ||||
| Block Design | X | X | X | |||||
| Digits Forward | X | X | X | X | X | |||
| Digits Backward | X | X | X | X | X | X | ||
| Information | X | X | X | |||||
| Digit Symbol | X | X | X | X | X | |||
| Cued Recall | X | X | X | |||||
| Free Recall | X | X | X | X | X | |||
| Category Fluency | X | X | X | X | X | |||
| Picture Memory | X | |||||||
| Figure Logic | X | |||||||
| Vocabulary | X | X | ||||||
| Boston Naming Test | X | X | ||||||
| Outcomes | ||||||||
| Progressive Matrices | X | |||||||
| Serial 7’s | X | |||||||
| Trail-Making Task | X | X | ||||||
| Card Rotation | ||||||||
| Age | X | X | X | X | X | X | X | X |
| Gender | X | X | X | X | X | X | X | X |
| Education | X | X | X | X | X | X | X | X |
| Race | X | X | X | X | ||||
| Ethnicity | X | X | X | X | ||||
| Marital Status | X | X | X | X | X | X | X | X |
| Personality | ||||||||
| Self-Rated Health | X | X | X | X | X | X | ||
| Heart Problems | X | X | X | X | X | X | X | X |
| Stroke | X | X | X | X | X | X | X | X |
| Diabetes | X | X | X | X | X | X | X | X |
| Cancer | X | X | X | X | X | X | X | X |
| Respiratory Problems | X | X | X | X | X | X | X | X |
| Smoking | X | X | X | X | X | X | X | |
| Alcohol | X | X | X | X | X | X | X | |
1.4.3 Table S3
Next, I create table S3, which indicates previous uses of the samples, their findings, and the distinction between them and the present study.
uses <- read_xlsx(destfile, sheet = "Table S3")
meas <- uses %>%
kable(.
, "html"
, caption = "<strong>Table S3</strong><br><em>List of Prior Publications Examining Personality-Dementia or Neuropathology Associations</em>"
, escape = F
, col.names = paste0("<strong>", colnames(uses), "</strong>")
, align = c("l", rep("c", 7), "l", "l")
) %>%
kable_classic(full_width = F, html_font = "Times New Roman")
meas| Paper | Year | Ref # | Sample(s) | N | Data | Measures | Outcomes | Findings | Comparison with Current Study |
|---|---|---|---|---|---|---|---|---|---|
| Terracciano et al. | 2014 | 5 | BLSA (IPD) | N = 1671 | IPD + meta-analysis | E, A, C, N, O | Cognitive Status |
N -> higher risk C -> lower risk |
BLSA not used in the present sample. Meta-analyzed results from Rush-MAP and ROS, but included much smaller sample (we used an additional 10-15 years of data). |
| Wilson et al. | 2006 | 50 | Rush-MAP | N = 648 | IPD | N | Cognitive Status | N -> higher risk |
Additional waves of data (~15 years of additional follow-ups). Used Cox Proportional Hazards Models. Only examined Neuroticism |
| Wilson et al. | 2007 | 6 | ROS | N = 997 | IPD | C | Cognitive Status | C -> lower risk |
Additional waves of data (~15 years of additional follow-ups). Used Cox Proportional Hazards Models. Only examined Conscientiousness |
| Wilson et al. | 2015 | – | ROS, Rush-MAP | N = 309 | IPD | C | NFT, Lewy bodies, chronic gross cerebral infarctions, and hippocampal sclerosis,, terminal decline | C -> slower terminal but not preterminal decline |
Additional waves of data. Focus was on cognitive decline, not diagnosis. Only examined deceased participants. Additional waves of follow-up data. |
| Terracciano et al. | 2017 | 49 | HRS | N = 13,882 | IPD | E, A, C, N, O | Cognitive Status |
N -> higher risk C, A -> lower risk |
Used Cox Proportional Hazards Models. Additional waves of follow-up data. |
| Yoneda et al. | 2020 | 52 | EAS LASA |
N (EAS) = 785 N (LASA) = 1300 |
IPD | E, A, C, N, O | Cognitive Status | N increases -> higher risk |
We additionally include neuropathology data from EAS. Examined associations between personality and change and cognitive status, not baseline levels. |
| Duchek et al. | 2020 | 34 | WUSM-MAP | N = 436 | IPD | N, C | In vivo neuropathology Clinical Dementia Ratings | C -> lower early dementia risk |
Focused on in vivo neuropathology, rather than neuropathology at autopsy. Only examined taransitions to early stage dementia. Only investigated N and C. Additional waves of follow-up data. |
| Graham et al. | 2021a | 51 | ROS Rush-MAP |
N (ROS) = 783 N (MAP) = 857 |
IPD | E, A (ROS), C, O (ROS), N | Cognitive Resilience (residual of global cognitive function / decline regressed on pathology) | N-> worse resilience |
Focused on association between personality traits and asymmetry between neuropathology and cognitive function / decline. Did not account for dementia diagnoses. Additional waves of follow-up data. |
| Aschwanden et al. | 2020 | – | ELSA HILDA |
N (ELSA) = 6,887 N (HILDA) = 2,778 |
Meta-analysis | E, A, C, N, O | Cognitive Status from cognitive tests | C -> lower dementia risk (ELSA only) |
Used Cox Proportional Hazards Models Different cognitive status indicator that allowed us to use more of the sample. Additional waves of follow-up data. |
| Aschwanden et al. | 2021 | 19 | Rush-MAP ROS WUSM-MAP ELSA HILDA |
N (Rush-MAP) = 648 N (ROS) = 904 N (WUSM-MAP) = 436 N (ELSA) = 6887 N (HILDA) = 2778 |
IPD | E, A, C, N, O | Cognitive Status |
N -> higher risk C -> lower risk |
Used Cox Proportional Hazards Models. Only meta-analyzed existing previous data that, in many cases, had many fewer waves of follow-up. Covariates determined by previous publications. |
| Graham et al. | 2021b | 22 | EAS MAP ROS SATSA |
N (EAS) = 737 N (Rush-MAP) = 1233 N (ROS) = 1466 N (SATSA) = 707 |
IPD | E, A, C, N, O | Cognitive Status | O -> post-dementia decline |
Additional waves of data for EAS, Rush-MAP, and ROS. Focus was on personality predictors of cognitive decline (slope) and cognitive decline following dementia diagnosis. No reported associations between personality traits and cognitive status. |