Examining the Association Between Midlife Obesity and Cognitive Impairment in a Longitudinal Population-Based Cohort Restricted; Files Only

Abstract

Introduction- Midlife obesity has been identified as a major modifiable risk factor for Alzheimer's Disease and related dementias (ADRD). However, the precise link between BMI and the onset of cognitive changes remains unclear. This study investigates this association within the Health and Retirement Study (HRS) cohort.

Methods- We analyzed data from 20,141 participants in the HRS cohort. Cognitive status was assessed using modified Telephone Interview for Cognitive Status (mTICS) scores, and BMI data was collected throughout the follow-up period. Logistic and linear regression models were employed to evaluate the association between BMI and the odds of developing cognitive impairment as well as the age of onset of cognitive impairment. Polygenic Risk Scores (PRS) for BMI were estimated for a subsample (n=6,667) of HRS participants with European ancestry and genotyping data. Logistic and linear models were then used to explore if PRS for BMI was associated with a higher risk or earlier onset of cognitive impairment.

Results- Among all HRS participants, when adjusting for education, age, and race, lower BMI tertile (BMI < 25.5) was associated with higher odds of cognitive impairment compared to the middle tertile (BMI 25.5-29.9) (OR=1.27; p= <0.001). The linear model showed a weak, negative association between increasing age of onset and lower BMIs (r2= 0.369), but a U-shaped distribution was seen where lower BMI was associated with both earlier and older onset ages of cognitive impairment using an overfitted Loess Regression. In our genotypic analysis, PRS of BMI were not associated with cognitive impairment or the age of onset of cognitive impairment.

Discussion- The relationship between BMI and the onset of cognitive impairment in the HRS population has been revealed to be complex, where the onset age of cognitive impairment does not have an explicitly linear relationship with BMI, and the odds of cognitive impairment are not always associated with BMI. BMI PRS did not reveal a genetic association to the odds of cognitive impairment, yet it may be important to include other genetic ancestries before ruling out the use of polygenic risk scoring in BMI. BMI may be one of the factors contributing to the pathogenesis of ADRD as nutritional intake and behavioral changes occur among adults over the age of 55.

Table of Contents

Table of Contents

Introduction..................................................................................................................................... 8

i. Obesity 8

ii. Cognitive Impairment and Alzheimer’s Disease 8

iii. Obesity as a Modifiable Risk Factor for AD 9

iv. Previous Studies Exploring the Association of Cognitive Impairment and BMI 11

v. Psychosocial factors leading to Lower Weight 12

vi. Polygenic Risk Scores of BMI 13

vii. Knowledge Gaps in the Field 15

Methods ........................................................................................................................................ 16

i. Study Sample: Health and Retirement Study (HRS) 16

ii. Body Mass Index (BMI) 17

iii. Race/Ethnicity 18

iv. Cognitive Impairment Status (mTICS Score) 18

v. Age to Cognitive Onset (only among those cognitively impaired) 19

vi. Polygenic Risk Scores of BMI (PRS) 19

Results ........................................................................................................................................... 22

Characteristics of the Overall HRS cohort (N= 20,305) 22

Table I. Sample Characteristics of the HRS Participants 1996-2016 .................................... 22

Characteristics of the Genotyped Cohort (N=6667) 24

Table 2. Sample Characteristics of the Genotyped Cohort 1996-2016 (N=6667) ................ 24

Full Multivariate Logistic Model shows an Association Between Increasing Average BMI

Scores and a small decrease in the odds of cognitive impairment among all HRS participants

(N=19862) 25

Table 3. Results of Logistic Model Examining the Association Between Increasing Average

BMI Scores and odds of cognitive impairment among all HRS participants (N=19862) ..... 25

Significant Association among HRS participants who fall into the lower BMI tertile and

Higher Odds of Cognitive Impairment, but not in the higher BMI tertile. 26

Figure 1. Boxplot Distribution of BMI average scores among the BMI tertiles ................... 26

Table 4. Logistic Model Using BMI Tertiles Revealed Increased Odds of Cognitive

Impairment among HRS participants classified in the lower BMI tertile group when

compared to the average BMI group. .................................................................................... 27

Stratification by Race when looking at Continuous BMI among HRS Participants showed

Significant associations between BMI and the odds of Cognitive Impairment among White,

Non-Hispanics 27

Table 5. .................................................................................................................................. 27

When BMI is reclassified into Tertiles, Stratification by Race Shows a Higher Overall Odds of

cognitive impairment in the Lower BMI tertile Group of White, Non-Hispanic Participants

28Table 6. Results of Logistic Models Examining the Association Between Classification in

extreme BMI Tertiles and Differential odds of cognitive impairment among all HRS

participants, stratified by Race. (N=19862)........................................................................... 29

HRS Participants with an Earlier Onset Age of Cognitive Impairment Had Higher BMI Scores,

but after the age of 60, lower average BMI Values are associated to Onset of CI (N= 3111)

29

Figure 2. Linear Association Between the Age of CI Onset and BMI .................................. 30

Attempting to visualize the U-shaped distribution Using a Loess Regression model, an

Overfitting Model 30

Figure 3. Loess Regression Model of Age of CI Onset and average BMI among all HRS

participants (N=3111) ............................................................................................................ 31

BMI Tertile Groups Showed Differential Cox Proportional Hazards In Relation to the Onset

Time to Onset of Cognitive Impairment in the HRS population (N=3111) 31

Figure 4. Cox Proportional Hazard model looking at the Association of Median-

Dichotomized BMI PRS estimates and time to onset of cognitive impairment (N=3111) ... 32

Polygenic Risk Scores (PRS) Relating to BMI and GWAS P-value threshold Optimization

Models resulted in BMI PRS explaining 10% of the variance in BMI (N=6677) 32

Figure 5. PRS model fit and GWAS P-value thresholds for associations between BMI PRS

and BMI continuous scores ................................................................................................... 33

Table 5. Results of Running the Default Parameters at the GWAS interval (5.0 x 10^-5) ... 34

Participants with higher BMI PRS Did Not Have A Higher Risk of Earlier Onset of Cognitive

Impairment (N=1214) 34

When PRS Scores were Dichotomized, the Cox Proportional Hazard Model Did Not Find

Differential Onset of Cognitive Impairment during the follow-up period, 1996-2016 34

Figure 6. Cox Proportional Hazard model looking at the Association of Median-

Dichotomized BMI PRS estimates and time to onset of cognitive impairment (N=1214) ... 35

Discussion ..................................................................................................................................... 36

Limitations 40

Clinical Significance 41

Conclusion .................................................................................................................................... 42

References ..................................................................................................................................... 44

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Epidemiology
Subfield / Discipline	Epidemiology - MPH & MSPH
Degree	M.P.H.
Submission	Master's Thesis
Language	English
Research Field	Health Sciences, Public Health Artificial Intelligence
Keyword	Health and Retirement Study BMI Polygenic Risk Scores mTICS Longitudinal Cohort Midlife Obesity
Committee Chair / Thesis Advisor	Anke Huels, Emory University
Committee Members	Aliza Wingo, Emory University Thomas Wingo, Emory University
Partnering Agencies	Emory University schools, faculty or affiliated programs

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	File download under embargo until 20 May 2026	2024-04-25 16:43:30 -0400	File download under embargo until 20 May 2026

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions