Association of type 2 diabetes mellitus with dementia‐related and non–dementia‐related mortality among postmenopausal women: A secondary competing risks analysis of the women's health initiative

Abstract

INTRODUCTION

Alzheimer's disease (AD) and AD‐related dementias (ADRD) are leading causes of death among older adults in the United States. Efforts to understand risk factors for prevention are needed.

METHODS

Participants (n = 146,166) enrolled in the Women's Health Initiative without AD at baseline were included. Diabetes status was ascertained from self‐reported questionnaires and deaths attributed to AD/ADRD from hospital, autopsy, and death records. Competing risk regression models were used to estimate the cause‐specific hazard ratios (HRs) and 95% confidence intervals (CIs) for the prospective association of type 2 diabetes mellitus (T2DM) with AD/ADRD and non‐AD/ADRD mortality.

RESULTS

There were 29,393 treated T2DM cases and 8628 AD/ADRD deaths during 21.6 (14.0–23.5) median (IQR) years of follow‐up. Fully adjusted HRs (95% CIs) of the association with T2DM were 2.94 (2.76–3.12) for AD/ADRD and 2.65 (2.60–2.71) for the competing risk of non‐AD/ADRD mortality.

DISCUSSION

T2DM is associated with AD/ADRD and non‐AD/ADRD mortality.

Highlights

• Type 2 diabetes mellitus is more strongly associated with Alzheimer's disease (AD)/AD and related dementias (ADRD) mortality compared to the competing risk of non‐AD/ADRD mortality among postmenopausal women.

• This relationship was consistent for AD and ADRD, respectively.

• This association is strongest among participants without obesity or hypertension and with younger age at baseline, higher diet quality, higher physical activity, higher alcohol consumption, and older age at the time of diagnosis of type 2 diabetes mellitus.

1. BACKGROUND

Alzheimer's disease (AD) is a chronic neurodegenerative disease that is a leading cause of dementia in older adults. A total of 6.5 million adults (10.7%) age 65 or older in the United States are estimated to have AD, ¹ nearly two‐thirds of whom are women. ² Due to the aging population of the United States, the prevalence of AD is estimated to grow to over 13.8 million by the year 2060. ² The increasing prevalence in the United States is similar to global estimates. ³ Since 2000, reported deaths from AD have increased by 146.2%, making AD the sixth leading cause of death in the United States. ⁴ By 2050, AD is projected to be the cause of 43% of all older adult deaths. ⁵ The economic impact of AD in the United States is staggering and growing. In 2022, the national cost of caring for people with AD and AD‐related dementias (ADRD), was estimated to be $321 billion, not including an additional $271 billion representing unpaid caregiving. ¹ By 2050, the national cost of caring for AD/ADRD is projected to triple to nearly $1 trillion. ¹

Due to the projected major burden to the health care establishment and the economic impact, efforts to better understand the prevention and treatment of AD/ADRD are urgently needed. Evidence suggests that modifiable risk factors such as the prevention of diabetes mellitus (DM) ⁶ could prevent up to 30% of AD cases worldwide. ⁷ It is unclear whether treatment of DM prevents AD/ADRD. ⁸ However, AD shares many molecular and cellular features with DM, ⁶ so much so that some researchers have referred to AD as “type 3 diabetes.” ⁹ Two meta‐analyses of observational studies found that DM is associated with a 53% increased risk of AD ¹⁰ and 60% increased risk of AD/ADRD ¹¹ ; however, the majority of the included studies are limited by small sample size, self‐reported disease status, retrospective analysis, and case‐control designs. In addition, a recent prospective analysis of the Women's Health Initiative (WHI) observational study found a strong association between DM and risk of AD ¹² ; however, this study also relied on self‐reported AD status as the outcome.

Historically AD/ADRD mortality has been severely under‐reported; for example, a study of the National Health and Nutrition Examination Survey (1999–2010) reported a null association of DM with AD mortality. ¹³ However, the study sample contained only 56 AD‐related deaths; therefore, the few outcome events likely contributed to the null finding. Since the early 2010s, increased awareness and billing practice changes have led to increased reporting of AD/ADRD on death certificates. ¹⁴ Therefore, prospective cohorts with linked mortality data collected after the increased reporting of AD/ADRD may allow for an opportunity to link modifiable risk factors to AD/ADRD mortality. Thus the objective of this study was to examine the prospective relationship between type 2 DM (T2DM) and risk of AD/ADRD mortality among the WHI cohort of postmenopausal women.

2. METHODS

2.1. Study population

The design, methods, recruitment, baseline characteristics, and reliability of data of the WHI have been published elsewhere. ¹⁵ , ¹⁶ , ¹⁷ Beginning in 1993, the WHI recruited 161,808 postmenopausal women between 50 and 79 years of age into clinical trials or an observational study. When the first phase of WHI ended in 2005, participants were invited to join subsequent WHI Extension studies, with 71.3% and 57.8% enrolling in extensions 1 and 2, respectively. Data from the WHI observational study and clinical trials were considered for this study. Participants with missing information (no response to questions) on diabetes (n = 928), with prevalent self‐reported (n = 88) or missing information on (n = 9069) AD at baseline, and participants with implausible energy intakes (<600 kcal/day or >5000 kcal/day; n = 4897) were excluded from the analysis. In addition, participants with prevalent diabetes at baseline who reported using only insulin therapy (n = 397) or being diagnosed prior to age 30 years (n = 263) were excluded to limit possible confounding of type 1 DM, resulting in a final sample size of 146,166 participants (Figure S1 ).

2.2. Assessment of exposures

Participants were asked if a physician had ever told them they had “sugar diabetes or high blood sugar” when they were not pregnant at baseline. If a participant responded in the affirmative, they were asked additional questions about treatment with insulin or oral antidiabetic medications. In this study, prevalent T2DM was defined as the subset who reported being treated. Incident T2DM was ascertained annually by self‐report of treatment for DM since their last health assessment; thus, only self‐reported treated incident T2DM can be ascertained. The accuracy of self‐reported treated DM in WHI has been found previously to be valid. ¹⁸

2.3. Ascertainment of outcomes

Deaths were ascertained by reviewing death certificates, medical records, autopsy reports, and by linkage to the National Death Index. ¹⁹ Death certificates and hospital records were obtained and adjudicated by adjudicators who were unaware of study components or randomization assignment. Deaths in the clinical trial component of the WHI were centrally adjudicated, whereas other deaths, including AD/ADRD, were adjudicated locally. ¹⁹ Records from the most relevant hospital admission preceding death and from the time of death, autopsy records, and the death certificate were used by adjudicators to determine the causes of death. For many deaths occurring out of the hospital, documentation was limited to the death certificate and records of the most recent admission to hospital before death. In these instances, the cause of death was determined from the death certificate, ¹⁹ which accounted for 30.0% of deaths in the present study. Ascertained deaths attributed to AD included International Classification of Diseases, Tenth Revision (ICD‐10) code G30 and those attributed to ADRD included ICD‐10 codes F01, F02, F03, F05, G30, G31, G32, and R62.

RESEARCH IN CONTEXT

1. Systematic review: Although the association between diabetes and Alzheimer's disease (AD) and AD‐related dementias (ADRD) is well known, few prospective cohort studies have investigated the association of diabetes with mortality due to AD/ADRD.

2. Interpretation: Type 2 diabetes mellitus (T2DM) is strongly associated with AD/ADRD and non‐AD/ADRD mortality in postmenopausal women. The results show that this association is strongest among participants without obesity or hypertension and with younger age at baseline, higher diet quality, higher physical activity, higher alcohol consumption, and older age at the time of T2DM diagnosis.

3. Future directions: Due to the increasing prevalence of AD/ADRD, there is an urgent need to identify modifiable risk factors. This study suggests that AD/ADRD mortality may be a useful outcome to evaluate risk factors for AD/ADRD in prospective cohorts with linked mortality data. Public health efforts to control and prevent T2DM may have the additional benefit of lowering the risk of death from AD/ADRD.

2.4. Potential confounders

Information on several covariates were included in fully adjusted models to evaluate the independent association of T2DM and AD/ADRD mortality. At baseline, participants self‐reported the following information: demographic characteristics (age, race, ethnicity, education, annual income, relationship status, health insurance), lifestyle factors (smoking status, recreational physical activity, alcohol intake, menopausal hormone therapy use, sleep disturbance, overall diet quality), and medical history (self‐reported treated hypercholesterolemia, self‐reported treated hypertension, and stroke prior to enrollment). Strokes during follow‐up were centrally adjudicated and ascertained from medical records. The 2015 healthy eating index was used to evaluate diet quality, where a higher score indicates better diet quality. ²⁰ Metabolic equivalent task hours per week (MET‐h/wk) of recreational physical activity for each participant were calculated for moderate to vigorous intensity recreational physical activity. ¹⁷ , ²¹ Study‐specific covariates including the WHI survey regions (Northeast, South, Midwest, and West) and the WHI components (clinical trial and observational study) were also included. Anthropometrics including weight, height, and systolic and diastolic blood pressure were measured during clinic visits using standard methods and were used to calculate body mass index (BMI) as weight (in kilograms) divided by height (in meters squared) and hypertension status defined as systolic ≥140 mmHg or diastolic ≥90 mmHg or self‐reported treatment for hypertension. Time‐varying hypertension status and BMI collected in nearest proximity to death were included in the models. The missing information of categorical variables was coded into their own category.

2.5. Statistical analysis

Study sample characteristics were summarized by diabetes status, and the chi‐square tests was used to compare categorical variables and analysis of variance (ANOVA) for continuous variables that are normally distributed or the Kruskal‐Wallis nonparametric test for continuous variables that are not normally distributed.

Cox competing risk regression models ²² were used to estimate cause‐specific hazard ratios (HRs) and 95% confidence intervals (CIs) for associations between T2DM and risk of AD/ADRD mortality accounting for the competing risk of non‐AD/ADRD mortality. Time‐to‐event was calculated as the number of days from the first instance of self‐reported incident T2DM or from enrollment date for participants with prevalent T2DM at baseline to the date of death. Time‐to‐event of participants who had not died or had withdrawn from the study before the end of follow‐up were censored. Covariates in fully adjusted models included baseline information on age, hypercholesterolemia, sleep disturbance, educational attainment, income, marital status, health insurance, race, ethnicity, smoking status, alcohol use, hormone therapy use, recreational physical activity, diet quality, WHI region, and WHI component. In addition, fully adjusted models include time‐varying information on BMI and hypertension that was updated throughout follow‐up to include information proximal to death and ever having a stroke.

To limit confounding due to pre‐observation disease processes, a sensitivity analysis was conducted excluding participants who died within the first 2 years of follow‐up. To limit confounding due to dementia potentially caused by previous stroke, a sensitivity analysis was conducted excluding participants who had a history of stroke prior to death. Stratified standard cox proportional hazards regression models were conducted to evaluate interactions from other potential risk factors for AD/ADRD including alcohol consumption, diet quality, recreational physical activity, smoking status, sleep disturbance, hypercholesterolemia status, hypertension status, obesity status, and age at baseline. To further evaluate how age modifies the risk of AD/ADRD mortality, we conducted a subgroup competing risk analysis including only participants with incident T2DM and we used categories of age at T2DM diagnosis (<65, 65–80, and 80+ years) as the exposure. To assess linear trends across diagnosis age categories, age at T2DM diagnosis was fit as a continuous variable in standard cox proportional hazards regression models.

All analyses were conducted with two‐sided tests (α = 0.05) in SAS version 9.4 (SAS Institute, USA).

3. RESULTS

There were 5941 treated T2DM cases prevalent at baseline and another 23,452 incident cases occurred for a total of 29,393 treated T2DM cases. Participants with T2DM were more likely to be younger, non‐drinkers, of Hispanic/Latino ethnicity, and have lower family income, lower educational attainment, lower diet quality, BMI  ≥30.0 kg/m², hypercholesterolemia, hypertension, history of stroke, and sleep disturbance (Table 1). In addition, participants with T2DM were less likely to be White, physically active, current menopausal hormone therapy users, married, and have health insurance.

TABLE 1.

Participant characteristics by type 2 diabetes mellitus status. ^a

Characteristics	Without T2DM	With T2DM b	p‐value
No. of participants	116,773	29,393
Age (years)	63.5 ± 7.29	62.8 ± 6.90	<0.0001
Race (%)
American Indian/Alaska Native	0.25	0.49	<0.0001
Asian	2.46	2.76
Native Hawaiian/Other Pacific Islander	0.08	0.12
Black	7.09	13.2
White	87.4	79.9
More than one race	1.07	1.62
Unknown/Not reported	1.67	1.96
Ethnicity (%)
Not Hispanic/Latino	95.3	94.0	<0.0001
Hispanic/Latino	3.90	5.15
Unknown/Not reported	0.78	0.86
Education (%)
High school or less	21.1	24.0	<0.0001
Some college	37.2	38.8
College	11.4	9.69
Postgraduate or professional	29.5	26.7
Missing	0.75	0.76
Income (%)
<$20,000	14.2	17.9	<0.0001
$20,000–$50,000	41.7	42.7
>$50,000	37.3	33.4
Missing	6.80	5.95
Health Insurance (%)
No	4.03	5.32	<0.0001
Yes	95.0	93.7
Missing	0.92	0.99
Relationship status (%)
Single	4.41	4.36	<0.0001
Divorced, widowed, or separated	32.3	34.8
Married or marriage‐like relationship	62.8	60.4
Missing	0.45	0.49
WHI region (%)
Northeast	22.8	23.0	<0.0001
South	25.1	26.6
Midwest	22.4	22.2
West	29.8	28.2
WHI component (%)
Clinical trial	39.1	46.3	<0.0001
Observation study	60.9	53.7
Menopausal hormone therapy (%)
Never	32.1	34.9	<0.0001
Past	22.4	24.3
Current	44.4	39.8
Missing	1.04	1.00
Smoking status (%)
Never	50.3	50.1	0.75
Past	41.9	41.9
Current	6.70	6.86
Missing	1.20	1.17
Alcohol (%)
Non‐drinker	26.8	36.4	<0.0001
Moderate	59.7	55.1
Heavy	13.0	8.04
Missing	0.48	0.53
Sleep disturbance (%)
No	75.7	73.8	<0.0001
Yes	24.3	26.2
Physical activity (MET‐h/wk) (%)
<10	52.0	60.6	<0.0001
≥10	47.8	39.1
Missing	0.26	0.21
HEI 2015	65.5 ± 10.4	64.2 ± 10.4	<0.0001
BMI (kg/m2) category (%)
<18.5	1.21	0.42	<0.0001
18.5–24.9	36.1	18.7
25.0–29.9	35.2	31.1
≥30.0	27.1	49.4
Missing	0.38	0.43
Hypertension (%)
No	56.0	41.8	<0.0001
Yes	44.0	58.2
Hypercholesterolemia (%)
No	87.4	81.0	<0.0001
Yes	12.6	19.0
Stroke ever (%)
No	94.9	92.5	<0.0001
Yes	5.11	7.55

Abbreviations: T2DM, type 2 diabetes mellitus.

^a Data are shown as mean ± standard error or percentages.

^b Includes both prevalent and incident T2DM.

During 21.6 (14.0–23.5) median (IQR) years follow‐up, 4308 AD and 4320 ADRD deaths occurred, for a total of 8628 combined AD/ADRD deaths. Another 55,956 deaths from all other, non‐AD/ADRD‐related, causes occurred. Mean age at death was 4.7 years higher for AD/ADRD mortality compared to non‐AD/ADRD mortality (p < 0.0001). In crude models, the association of T2DM with risk of AD/ADRD mortality was significant with a cause‐specific HR (95% CI) of 2.27 (2.14, 2.41) but was lower than for the competing risk of non‐AD/ADRD mortality with a cause‐specific HR (95% CI) of 2.51 (2.46, 2.57; Table 2). In models adjusted for age, race, and ethnicity, the cause‐specific HR (95% CI) for the association of T2DM with risk of AD/ADRD mortality was 2.82 (2.66, 2.99), which was similar to the cause‐specific HR (95% CI) for non‐AD/ADRD mortality of 2.83 (2.77, 2.89). The association remained significant after additional adjustment for education, family income level, health insurance coverage, and marital status with a cause‐specific HR (95% CI) for AD/ADRD mortality of 2.82 (2.66, 2.99). With further addition of several lifestyle covariates including menopausal hormone therapy use, smoking status, alcohol intake, sleep disturbance, recreational physical activity, and diet quality, the cause‐specific HR (95% CI) for AD/ADRD mortality was 2.79 (2.63, 2.97). Finally, the association remained significant after further adjustment for BMI and comorbidities including hypercholesterolemia, hypertension, and history of stroke with a cause‐specific HR (95% CI) of 2.94 (2.76, 3.12), and was higher than the cause‐specific HR (95% CI) for non‐AD/ADRD mortality of 2.65 (2.60, 2.71).

TABLE 2.

Competing risks analysis for the association of type 2 diabetes mellitus with dementia‐related and non–dementia‐related mortality among postmenopausal women.

	Cause‐specific HRs (95% CIs)
	Without T2DM	With T2DM	Without T2DM	With T2DM
	AD		Non‐AD
No. of death/person‐years	3667/65173	641/7172	47647/731816	12629/125113
Crude model	1 (reference)	2.09 (1.92, 2.28)	1 (reference)	2.51 (2.46, 2.56)
Model 1	1 (reference)	2.59 (2.37, 2.82)	1 (reference)	2.84 (2.79, 2.90)
Model 2	1 (reference)	2.62 (2.40, 2.86)	1 (reference)	2.80 (2.75, 2.86)
Model 3	1 (reference)	2.61 (2.39, 2.85)	1 (reference)	2.77 (2.72, 2.83)
Model 4	1 (reference)	2.80 (2.56, 3.06)	1 (reference)	2.68 (2.63, 2.74)

	ADRD		Non‐ADRD
No. of death/person‐years	3578/63697	742/8172	47736/733292	12528/124113
Crude model	1 (reference)	2.45 (2.26, 2.65)	1 (reference)	2.49 (2.44, 2.54)
Model 1	1 (reference)	3.06 (2.82, 3.32)	1 (reference)	2.82 (2.76, 2.87)
Model 2	1 (reference)	3.02 (2.78, 3.27)	1 (reference)	2.78 (2.73, 2.84)
Model 3	1 (reference)	2.97 (2.74, 3.23)	1 (reference)	2.75 (2.70, 2.81)
Model 4	1 (reference)	3.06 (2.82, 3.33)	1 (reference)	2.67 (2.61, 2.73)

	AD/ADRD		Non‐AD/ADRD
No. of death/person‐years	7245/128869	1383/15344	44069/668120	11887/116941
Crude model	1 (reference)	2.27 (2.14, 2.41)	1 (reference)	2.51 (2.46, 2.57)
Model 1	1 (reference)	2.82 (2.66, 2.99)	1 (reference)	2.83 (2.77, 2.89)
Model 2	1 (reference)	2.82 (2.66, 2.99)	1 (reference)	2.79 (2.73, 2.85)
Model 3	1 (reference)	2.79 (2.63, 2.97)	1 (reference)	2.76 (2.70, 2.82)
Model 4	1 (reference)	2.94 (2.76, 3.12)	1 (reference)	2.65 (2.60, 2.71)

Abbreviations: AD, Alzheimer's disease; ADRD, Alzheimer's disease and related dementias; T2DM, type 2 diabetes mellitus; WHI, Women's Health Initiative.

Model 1: adjusted for age, race, and ethnicity.

Model 2: model 1 + education, family income level, health insurance, relationship status, WHI survey region, and WHI component.

Model 3: model 2 + menopausal hormone therapy use, smoking status, alcohol intake, sleep disturbance, recreational physical activity, and diet quality.

Model 4: model 3 + body mass index, hypertension, hypercholesterolemia, and stroke ever.

T2DM was also independently associated with an increased risk of AD and ADRD mortalities individually. For AD mortality, T2DM was associated with a cause‐specific HR (95% CI) of 2.80 (2.56, 3.06) compared to 2.68 (2.63, 2.74) for the competing risk of non‐AD mortality in fully adjusted models (Table 2). For ADRD mortality, T2DM was associated with a cause‐specific HR (95% CI) of 3.06 (2.82, 3.33) compared to 2.67 (2.61, 2.73) for the competing risk of non‐ADRD mortality in fully adjusted models.

After stratification by prevalent or incident T2DM status, the association of T2DM with AD/ADRD mortalities remained significant for both groups. After exclusion of participants with prevalent T2DM at baseline, the cause‐specific HRs (95% CIs) were 3.29 (2.98, 3.63) for AD mortality, 3.65 (3.33, 4.01) for ADRD mortality, 3.47 (3.25, 3.72) for AD/ADRD mortality, and 2.73 (2.66, 2.80) for non‐AD/ADRD mortality in fully adjusted models (Table 3). After exclusion of participants with incident T2DM, the cause‐specific HRs (95% CIs) were 1.91 (1.61, 2.26) for AD mortality, 2.05 (1.75, 2.38) for ADRD mortality, 1.98 (1.77, 2.22) for AD/ADRD mortality, and 2.58 (2.49, 2.67) for non‐AD/ADRD mortality in fully adjusted models.

TABLE 3.

Competing risks analysis for the association of type 2 diabetes mellitus with dementia‐related mortality and non–dementia‐related among postmenopausal women stratified by prevalent and incident T2DM. ^a

	Without T2DM		With T2DM
	No. of deaths/ person‐years	Cause‐specific HR (95% CI)	No. of deaths/person‐years	Cause‐specific HR (95% CI)
AD
Prevalent	3667/65,173	1 (reference)	152/2282	1.91 (1.61, 2.26)
Incident	3667/65,173	1 (reference)	489/4890	3.29 (2.98, 3.63)
ADRD
Prevalent	3578/63,697	1 (reference)	185/2791	2.05 (1.75, 2.38)
Incident	3578/63,697	1 (reference)	557/5382	3.65 (3.33, 4.01)
AD/ADRD
Prevalent	7245/128,869	1 (reference)	337/5073	1.98 (1.77, 2.22)
Incident	7245/128,869	1 (reference)	1046/10,271	3.47 (3.25, 3.72)
Non‐AD/ADRD b
Prevalent	44,069/668,120	1 (reference)	3883/45,074	2.58 (2.49, 2.67)
Incident	44,069/668,120	1 (reference)	8004/71,867	2.73 (2.66, 2.80)

Abbreviations: AD, Alzheimer's disease; ADRD, Alzheimer's disease and related dementias; T2DM, type 2 diabetes mellitus.

^a Models are adjusted for all factors in Model 4.

^b Non‐AD/ADRD mortality HRs are from competing risk analysis with AD/ADRD mortality.

For AD/ADRD mortality, results of fully adjusted stratified analyses were similar to the primary analysis and there were significant interactions for age at baseline, age of T2DM diagnosis, diet quality, recreational physical activity level, alcohol consumption, hypertension status, and obesity status (p ≤ 0.03 for all; Tables S1–S7) but not for smoking status, sleep disturbance, or hypercholesterolemia (Tables S8–S10). Stronger associations of T2DM with AD/ADRD mortality were observed for participants with younger baseline age, older age at T2DM diagnosis, higher diet quality, more recreational physical activity, heavy alcohol consumption, without hypertension, and without obesity.

4. DISCUSSION

Among postmenopausal women, T2DM was associated with increased risk of AD/ADRD mortality. The associations remained significant after adjustment for demographic, socioeconomic, lifestyle, and comorbidity risk factors for AD/ADRD mortality.

The results of this prospective cohort study are generally consistent with the previous findings on the association of DM with AD and AD/ADRD ¹⁰ , ¹¹ ; however, few studies have evaluated the prospective association of T2DM with AD/ADRD mortality or modifying factors. An analysis of the Clinical Practice Research Datalink found that death attributed to AD/ADRD increased from 0.9 to 5.1 deaths/1000 person years between 2001 and 2018 among people with DM in the United Kingdom. ²³ However, a similar analysis of the National Health Interview Survey found that AD mortality did not change among people with DM in the United States between 1990 and 2015. ²⁴ Similarly, one study of 15,513 participants in the National Health and Nutrition Examination Survey (1999–2010), found no association between DM and risk of AD mortality in the United States. ¹³ It should be noted that with regard to the two studies in the United States, most of the data analyzed was collected prior to the increased awareness and billing practice changes that have led to improved reporting of AD/ADRD deaths on death certificates. ¹⁴

This change in billing practice may partially explain why participants who were younger at baseline were at increased risk of AD/ADRD mortality compared to older participants at baseline. A higher proportion of younger participants in WHI likely died after these changes were implemented and were more likely to have death attributed to AD/ADRD. Age is one of the greatest risk factors for AD/ADRD ¹ ; thus the results of the analysis in the present study stratified by baseline age seemingly contrast with this well‐known risk factor. Therefore, to further explore how age modifies the risk of AD/ADRD mortality, we conducted a subgroup analysis restricted only to participants who developed T2DM during follow‐up to evaluate how the age at T2DM diagnosis modifies the risk for AD/ADRD mortality. The results of this analysis suggest that older age at T2DM diagnosis is associated with increased risk of AD/ADRD mortality. This finding suggests that emphasis on T2DM prevention among older adults may prevent the burden of AD/ADRD mortality; however, this finding may also be influenced by the billing practice changes from the early 2010s. ¹⁴ Participants who were diagnosed with T2DM at an older age (i.e., a larger proportion near the billing practice changes for AD/ADRD) may have been more likely to have their death attributed to AD/ADRD. Future prospective studies are needed to confirm this finding.

Due to the projected burden of AD/ADRD on the economy and health care system in the United States, identification of and public health action on modifiable risk factors to prevent AD/ADRD are urgently needed. In the United States, DM affects 37.3 million adults (11.3% of the adult population), ²⁵ and one study estimates the prevalence to increase to 60.6 million by the year 2060. ²⁶ Although alarming, the projected increase in DM prevalence may represent a public health opportunity, as a healthy diet and lifestyle may prevent up to 90% of T2DM cases. ²⁷ Decreasing the prevalence of T2DM over time is suggested by some, but not all, ²⁸ modeling studies to significantly decrease the future burden of AD. ²⁹ Aside from the link between T2DM and AD/ADRD, public health interventions to promote healthy diets and lifestyle may independently reduce risk of AD/ADRD. ³⁰ Notably, a recent study found that higher adherence to the Mediterranean and MIND (Mediterranean – DASH Diet Intervention for Neurogenerative Delay) diets were associated with less postmortem amyloid beta load among cadaver brains with AD pathology. ³¹

The interaction between alcohol consumption and T2DM regarding risk of AD/ADRD mortality observed in this study is striking. This study found that the association between T2DM and AD/ADRD mortality is strongest among participants with the highest alcohol consumption and lowest among participants who consumed no alcohol. This finding corroborates findings from two meta‐analyses of observational studies that showed that high alcohol consumption is associated with increased risk of AD and AD/ADRD. ³² , ³³ However, both studies observed that low‐to‐moderate alcohol consumption is associated with a reduced risk of AD and AD/ADRD, ³² , ³³ which contrasts with the results of the present study.

Because AD/ADRD is an age‐related condition, women with healthier lifestyles (e.g., higher diet quality and more physical activity) may live longer and, therefore, be at higher risk of AD/ADRD mortality. The differences observed in the risk of AD/ADRD mortality in analyses stratified by diet quality, recreational physical activity, obesity, and hypertension in the present study are possible evidence of this issue. Alternatively, women with low diet quality, low recreational physical activity, obesity, or hypertension may be more likely to die from non‐AD/ADRD causes. Because people with AD/ADRD have a higher risk for several comorbid conditions, ³⁴ , ³⁵ under‐reporting of AD/ADRD mortality on death certificates due to misclassification of the outcome is possible. ³⁶ Regardless of the reason, misclassification of the outcome in the present study would have an effect toward the null; therefore, the results from this study may be lower than the true underlying relation of T2DM to AD/ADRD mortality.

This study is strengthened by the large sample size, prospective study design, and rich data collected in the WHI studies. However, this study also has several limitations. First, because of the adjudication of AD/ADRD mortality from death certificates, misclassification of the outcome is probable. Second, although shown to be valid, ¹⁸ T2DM status in this study was determined by self‐reported treatment. Therefore, we cannot rule out potential misclassification of the exposure in our analysis. Third, despite the rich data collected in the WHI studies, data on all risk factors for AD/ADRD such as apolipoprotein E (APOE) genotype ³⁷ are available only for small subsets of the study sample. Therefore, as in all observational studies, residual confounding may be unaccounted for due to other confounders not collected in the WHI or included in our models.

The association of T2DM with AD/ADRD has been observed in several studies; however, the majority have been limited by self‐reported AD, small sample size, retrospective analysis, and case‐control designs. ¹⁰ The findings from the present study represent one of the first prospective cohort studies with large sample size to investigate the association of T2DM with AD/ADRD mortality. The results of the present study suggest that public health efforts to prevent T2DM among women may also prevent AD/ADRD mortality.

AUTHOR CONTRIBUTIONS

T.J.T. designed research with supervision by W.B. T.J.T. conducted research, analyzed data, and wrote the manuscript. All authors contributed to the acquisition, analysis, or interpretation of the data, and revised the manuscript for important intellectual content. T.J.T. has primary responsibility for the final content and is the guarantor of the study. All authors read and approved the final manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

CONFLICT OF INTEREST STATEMENT

T.L.W. has equity interest in the following companies: Dr Terry Wahls LLC, TZ Press LLC, The Wahls Institute, PLC, FBB Biomed Inc, Foogal Inc, and the website http://www.terrywahls.com. She has financial relationships with BioCeuticals, MCG Health LLC, Genova Diagnostics, Levels Health Inc, Standard Process Inc, Vibrant America LLC, and the Institute for Functional Medicine. All other authors report no conflicts of interest with this work. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

The Women's Health Initiative (WHI) project was reviewed and approved by the Fred Hutchinson Cancer Research Center (Fred Hutch) Institutional Review Board (IRB) in accordance with the U.S. Department of Health and Human Services regulations at 45 CFR 46 (approval number: IR# 3467‐EXT). Participants provided written informed consent to participate. Additional consent to review medical records was obtained through signed written consent. Fred Hutch has an approved Federalwide Assurance on file with the Office for Human Research Protections (OHRP) under assurance number 0001920.

The present analysis was determined not human subjects research by the University of Iowa IRB (# 202303594) due to the use of de‐identified data for this secondary analysis.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Titcomb TJ, Richey P, Casanova R, et al. Association of type 2 diabetes mellitus with dementia‐related and non–dementia‐related mortality among postmenopausal women: A secondary competing risks analysis of the women's health initiative. Alzheimer's Dement. 2024;20:234–242. 10.1002/alz.13416