https://orcid.org/0000-0001-9611-742X
Figures
Abstract
Objective
Accumulating evidence suggests the possibility that early life exposures may contribute to risk of Alzheimer’s Disease (AD). This paper explores geographic disparities in AD mortality based on both state of residence in older age as well as state of birth measures in order to assess the relative importance of these factors.
Methods
We use a subset of a large survey, the NIH-AARP Diet and Health Study, of over 150,000 individuals aged 65–70 with 15 years of mortality follow-up, allowing us to study over 1050 cases of AD mortality. We use multi-level logistic regression, where individuals are nested within states of residence and/or states of birth, to assess the contributions of place to AD mortality variation.
Results
We show that state of birth explains a modest amount of variation in AD mortality, approximately 4%, which is consistent with life course theories that suggest that early life conditions can produce old age health disparities. However, we also show that nearly all of the variation from state of birth is explained by state of residence in old age.
Citation: Topping M, Kim J, Fletcher J (2021) Geographic variation in Alzheimer’s disease mortality. PLoS ONE 16(7): e0254174. https://doi.org/10.1371/journal.pone.0254174
Editor: Y. Zhan, German Centre for Neurodegenerative Diseases Site Munich: Deutsches Zentrum fur Neurodegenerative Erkrankungen Standort Munchen, GERMANY
Received: February 1, 2021; Accepted: June 22, 2021; Published: July 1, 2021
Copyright: © 2021 Topping et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The datasets used in this analysis were obtained from the National Cancer Institute. The data underlying the results presented in the study are available upon submitting a proposal, to be approved by the NIH-AARP Steering Committee at (https://www.nihaarpstars.com). Interested researchers must register with the NIH-AARP Diet & Health Study Tracking and Review System (STaRS) to submit a formal data request to STaRS. Interested researchers would be able to access data from the NIH-AARP Diet and Health Study (https://dietandhealth.cancer.gov) in the same manner as the authors. The authors did not have any special access privileges that others would not have.
Funding: This research was supported by the NIA (RF1AG062765). The authors gratefully acknowledge use of the facilities of the Center for Demography of Health and Aging at the University of Wisconsin-Madison, funded by NIA Center Grant P30 AG017266.
Competing interests: The authors declare that they have no competing interests.
Introduction
Place-based differences in later life mortality and health are a well-documented and increasing body of literature [1,2]. While life expectancy has improved for decades, there still remains a gap of 7.5 years across states [3]. Existing research has predominantly focused on individual level factors in health outcomes but have increasingly included information on the influence of place [4]. Research has found that an individual’s place of birth is connected with the development of chronic illnesses later in life such as cancer, diabetes, and heart disease [5–7]. Contextual factors also are considered as principal indicators of population health, but when and how these contexts come into play is not fully understood. Additionally, little is known on how early environmental factors affect mortality later in life.
The Developmental Origins of Adult Heath and Disease (DOHaD) and life course frameworks state that health disparities in later life could be a result of place-based exposures throughout the lifespan, not just current exposures [8]. The impact that these life exposures have can both accumulate into later life and influence other health outcomes [9–11]. This accumulation of risk across the life course, rather than at a particular stage, may be what is ultimately detrimental for individual’s health and well-being [12].
The past and present exposures individuals are subject to can be distinguished by migration. About one third of individuals in the United States live in a state in which they were not born in [13]. Therefore, a significant portion of the inequalities in health and mortality might be linked to exposures that people encountered early in life, opposed to a later period in the life course. Nevertheless, since mortality is often examined through the lens of the late period of one’s life, little consideration is given to the exposures early in an individual’s lifespan.
Past research has consistently documented key social conditions of health that persist throughout the life course and impact individual mortality risk [14]. Looking at the specific effects that state residence contexts have on mortality and health in adulthood, they typically examine a myriad of predictors, such as socioeconomic status, policy, social capital, and income inequality [15,16]. Conversely, other studies have examined the effects of state of birth on health outcomes in early life. State-level policies on poverty reduction and health care coverage have been shown to have an impact on infant health, such as birth weight [17,18]. Such poor health outcomes in early life have the potential to leave long lasting effects on lifetime health [19,20]. Moreover, unhealthy environments in early life typically can lead to challenges in learning, being employed, or have earning capability [21,22]. In turn, these difficulties undoubtedly have significant ramifications for health, well-being, and mortality in later life.
One poor health outcome that individuals encounter later in life is Alzheimer’s disease (AD). AD is a debilitating disease which shortens life expectancy, impairs memory, and is a key cause of physical disability and lower quality of life to older adults [23–25]. Since occurrence of AD is associated with an increase in age [26], it is presumed that the disease will have a higher prevalence due to greater population aging in society. Thus, there is a fundamental need to attempt to ascertain the determinants of the disease that is the sixth leading cause of death in the United States [27].
However, despite the prevalence of AD, few studies have attempted to connect early life exposures with mortality from the disease later in life. This is because a great deal of mortality studies in the United States often lack the ability to consider exposures at different points throughout the life course. Moreover, AD mortality studies often examine the period from the diagnosis to the point of death because it is a progressive disorder [24]. Related research focuses on either demographic or clinical factors that show a higher likelihood of mortality of those who already have the disease [28,29].
AD prevalence and mortality have been associated with a growing set of factors: race/ethnicity [25,30], educational attainment [31], nutrition [32], mental health [33], and air pollution [34,35]. While all of the factors outlined are indeed significant determinants of health, there is little literature on early life contexts and their impact on AD mortality. Moreover, there is no existing research regarding the association of one’s state of birth with AD mortality. As a crude measure of early life exposure, state of birth could provide an omnibus test of whether variation in early environments across the US are consequential for developing AD later in life.
This study is unique in the sense that it takes advantage of data that can link information regarding state of birth and state of residence for a large and representative sample of individuals in the United States. Most other research is limited to smaller convenience samples that are not well powered to examine state-level variation in AD mortality. Specifically, this paper addresses the following: (1) the impact that early environments and contexts contribute to differences in AD mortality and (2) what impact state of birth effects differ by social categories such as race/ethnicity, sex, and educational attainment. Ultimately, the aim of this paper is to increase the understanding of the critical determinants of health and health behaviors and effect that early environments have regarding outcomes throughout the life course.
Materials and methods
Data
This data utilized in this research was taken from the NIH-AARP Diet and Health Study (DHS). The DHS is a large prospective cohort from members of the American Association of Retired Persons (AARP), ranging from 50 to 71 years old, who responded to a mailed questionnaire between 1995 and 1996 [36]. Initially, over 3.5 million members of the AARP were mailed the questionnaire, resulting in 620,000 responses. Of those responses, nearly 570,000 provided data that was usable for analysis. The participants of this study were from six states (California, Florida, Louisiana, New Jersey, North Carolina, and Pennsylvania) and from two cities (Atlanta, Georgia and Detroit, Michigan) in the US. All cohort participants signed a written informed consent at enrollment, and the study protocol was approved by Special Studies Institutional Review Board of the National Cancer Institute.
DHS asked a comprehensive questionnaire which assessed lifestyle factors and diet of the participants at baseline. Specifically, the questionnaire addressed information on nutrient intake, along with health questions, family illness history, and other health-related conditions. Demographic information was collected from participants as well, such race/ethnicity, sex, and educational attainment, along with other variables to measure health outcomes and well-being. This source of data is unique due to its large sample sizes, which is essential in order to examine the rare outcome of AD mortality. Moreover, it is necessary to have large sample sizes in order to appropriately examine state-level variation. Other datasets (e.g., Health and Retirement Study) that include the variables laid out in this study are rather limited, due to the much smaller number of cases of AD mortality.
Starting from the initial 566,398 respondents in the original study, specific observations were dropped. First, of the original sample, we dropped 165,917 with invalid states of birth, or those born in United States territories (American Samoa, Guam, Puerto Rico, and the Virgin Islands), and those with missing values. Specifically, the observations dropped with missing values were about missing social security numbers, which would have prevented state identification in our analysis. Next, in order to focus on individuals at higher risk of experiencing AD mortality over a 15-year follow-up, a further 248,108 observations were dropped so that the sample is of individuals between the ages of 65 and 70. The final sample that was used for statistical analysis was 152,373 (Fig 1). We found that older adults are more likely to be in the analytic sample and those of other race groups are less likely to be in the analytic sample than non-Hispanic whites (S1 Table). Despite this, our final sample size provides more statistical power than alternative datasets available like the Health and Retirement Study, with about five times the amount of observations to examine [37].
Measures
AD mortality.
The key outcome variable in this study is AD mortality, which is ascertained from the follow-up of the DHS. The vital status of the individual was obtained by the annual linkage of the cohort in the DHS to the Social Security Administration Death Master Files in the US verification of vital status [38]. Cause of death information will then come from follow-up searches of the National Death Index, focused on death from AD (ICD-10 Code G30). Our data is linked to the NDI through 2011, allowing 15 years of follow up.
State of birth.
State of birth was created by using the first three digits of the individual social security number of each in the cohort (asked in initial survey), allowing information by state and year to be obtained [39].
State of residence.
State of residence was ascertained from the original survey, which asked the participants which state they lived in at the time. The eight states of residence reflected in this study are: California, Florida, Georgia, Louisiana, Michigan, New Jersey, North Carolina, and Pennsylvania.
Covariates.
Sex was used as a variable to account for the difference in mortality risk that exists between men and women. Likewise, race and ethnicity (non-Hispanic white, black and others) were utilized to account for the differential risk that exists along the boundaries of race and ethnicity. Educational attainment levels (less than high school, high school, some college, and college grade or more) were included to account for the influence that schooling provides throughout the life course with regard to health outcomes. Age is included as a covariate as well, specifically those that are between the ages of 65 and 70 years old to see the differences in health outcomes are impacted later in the life course of individuals. It is worth noting that we did not control for personal characteristics, such as physical and mental health status and functional status, because those characteristics are likely mechanisms, not confounders of state of birth.
Analytic strategy
In this study, a series of multi-level logistic regression models are used with state of residence and/or state of birth random effects. The full model is specified as the following equation:
where
and
. In the model, b0 is the average log-odds of mortality among individuals in the sample, εk is the effect of state of birth and εj is the effect of state of residence. Both are assumed to be normally distributed. εijk is the error term at the individual level. We control for a limited set of factors in order to not “over control” for the effects of state of birth on later life AD mortality. Our age control removes composition differences between the states.
The first and second model in the series includes state of birth random effects, with the second of the two incorporating all fixed effects. Then, the third and fourth run identical analyses, but with state of residence random effects, with the fourth including all fixed effects. The fifth and final model analyzes all fixed and random effects, in order to see differences between state of residence and state of birth variations.
The main results are first presented and include analysis from all states. Supplementary tables provide subsequent results that display the same models but exclude each individual state of residence to demonstrate the robustness of the models. Other models focus specifically on sex, race/ethnicity, and education to explore reductions in cross-state variation. These analyses are stratified in order to explore heterogeneity that may exist in the variation. Finally, two results tables are presented that exclude states which account for small portions of the sample. The first excludes states that make up less than one percent of the sample, and the second shows those with less than five percent. This is done to see if it is consequential for state of birth random effects, for they could be difficult to estimate with very small sample sizes.
In this study, Stata software version 16.1 (College Station, TX) were used for all statistical analyses. Likelihood Ratio (LR) tests were used to determine significance. All statistical tests were two-tailed, with the null hypothesis of no difference being rejected if p < 0.05.
Results
Descriptive statistics of the study sample are presented in Table 1. Among the 152,373 individuals in the sample, all states, save for Alaska, are accounted for with regard to state of birth. Approximately 55,142 (36.2%) were women and 97231 (63.8%) were men, and mean age among the sample was 67.4 (SD = 1.672). Over 90% of individuals were non-Hispanic white, with those who were not non-Hispanic whites accounting for 6.6%. For education, over one-third (38.7%) had a college education or greater, while those with some college, high school, or less than high school accounted for 21.8%, 29.6%, and 7.1%, respectively.
Table 2 presents the rate of concordance/discordance in state of birth and residence among our sample. For the pooled sample, about a quarter of the sample (22.79%) lived in a state that they were not born in. The discordance differs greatly by state of residence, ranging from 4.26% for California to 47.24% for Georgia.
The main results for all states are presented in Table 3, in the form of multi-level regression models that include both fixed and random effects in each model. Diagnostic statistics would indicate that the final model, which includes random effects of both state of birth and residence, performs best. Models 1 and 2 show that state of birth accounts for approximately 4.5% of the variance in AD mortality in the data, which is similar to results from Xu et al. (in press) for all-cause mortality using alternative data. Regarding the fixed effects, ages 65 through 68 are statistically significant across all models they are included in (models 2, 4, and 5). Meanwhile, for race/ethnicity, it shows that non-Hispanic blacks are significant across all models, whereas other racial groups are not. Models 3 and 4 then show that state of residence accounts for more than the state of birth models, at approximately 7.5% of the variation in AD mortality. However, since state of birth and state of residence are highly correlated, we estimate a final set of models that includes both random effects.
Our final models estimate the variance of state of residence in this model to be 7.6% and show that variation from state of birth can no longer be estimated. We interpret this finding to mean that a larger proportion of the variance in the risk of AD mortality could be explained by state of residence random effects, as opposed to state of birth.
In addition to our main results, we also conducted robustness and heterogeneity analyses. Analysis stratified by social determinants of health such as sex, race/ethnicity, and education show that there is limited heterogeneity in terms of how much variation there is between state of birth and state of residence (See S2–S4 Tables). Regarding sex and race/ethnicity, analysis showed that state of residence was the predominant factor in the risk of AD mortality, similar to the main results. Education, particularly those with some college education, shows that state of birth does make up a larger share of the variance; however, this is the only instance of this occurring in all result models. For other education levels, only college education or more is responsible for some of the variation (14%) in AD mortality.
S5–S12 Tables show that our results are not driven by any particular state of residence in the analysis and are similar for sociodemographic groups. S13 and S14 Tables exclude states of birth with small numbers of respondents and show the results are unchanged. In S15 Table, we also find that the results are robust to excluding those who lived in the state where they were born.
Discussion
Previous research has made significant contributions to the literature surrounding place-based differences in later life mortality. Despite this, much of this research has put greater emphasis on individual level factors and contemporaneous (i.e. old age) geographic contexts [4,5]. The aim of this study was to test the extent of the importance of broad-based early environments and contexts in accounting for variation in AD mortality. The analyses in this paper show that state of residence random effects explained the majority of geographic variance in the risk of AD mortality. This was the case across the main results and in supplementary robustness checks. An explanation for this could be that, while state of birth may explain health outcomes in later life as indicated in previous studies [9,15], they may not explain outcomes regarding AD mortality specifically.
These findings do partially support previous research done on how AD mortality is influenced by an individual’s level of educational attainment [31]. Further variables were to be included in the analysis, primarily those on health behaviors, but initial results suggested no need to examine those factors.
There are important limitations in this study. First, the data used for analysis only includes eight states of residence (California, Florida, Georgia, Louisiana, Michigan, New Jersey, North Carolina, and Pennsylvania). This limits the generalizability of our findings. Results from supplemental analyses suggest that no single state was responsible for the results. Future studies would benefit by employing larger sample sizes and additional states of residence. Another limitation is that our measure of early life context is state of birth rather than place of childhood. We also cannot estimate the impacts of durations of place-based exposures. Building from this, given that many people may move internally in their state of residence [13], it is difficult to ascertain the exposures that individuals faced without knowing if or when they moved. Thus, focusing on the state as the geographic unit of analysis will mask disparities within states, particularly those in which people were born and continued to reside in throughout their lives. Future studies should also examine inter-state heterogeneity to further document life course effects of exposures. A third limitation is the measurement bias of AD mortality. Many studies often use death certificates for the study of AD mortality, but often suffer from an underreporting of death due to uncertainties regarding coding practices in the cause of death [40,41]. However, when it comes to measurement of AD mortality, there is no agreed upon gold standard for measurement.
Conclusion
This study is one of the first studies to examine the independent and joint associations of state of birth and state of residence with AD mortality. Results point to state of residence having a larger role in geographic disparities that exist in AD mortality later in life. These findings show that there should be a greater focus on contemporaneous exposures, compared to early life ones to fully ascertain what influences the likelihood death from AD. Moreover, the boarder implications of this study are that specific conditions are better explained through a place of birth lens, such as diabetes, heart disease, or cancer [5–7], while illnesses such as AD are best looked at by place of residence, in this study at least. In short, this research highlights the importance of later life factors in the examination of degenerative chronic illnesses that disproportionately impact individuals towards the end of the life span.
Supporting information
S13 Table. Robustness: Excluding states of birth with less than 1% of the total sample.
https://doi.org/10.1371/journal.pone.0254174.s013
(DOCX)
S14 Table. Robustness: Excluding states of birth with less than 5% of the total sample.
https://doi.org/10.1371/journal.pone.0254174.s014
(DOCX)
S15 Table. Robustness: Excluding those who lived in the state where they were born.
https://doi.org/10.1371/journal.pone.0254174.s015
(DOCX)
Acknowledgments
Declarations
This research was supported by the Intramural Research Program of the NIH, National Cancer Institute. Cancer incidence data from the Atlanta metropolitan area were collected by the Georgia Center for Cancer Statistics, Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia. Cancer incidence data from California were collected by the California Cancer Registry, California Department of Public Health’s Cancer Surveillance and Research Branch, Sacramento, California. Cancer incidence data from the Detroit metropolitan area were collected by the Michigan Cancer Surveillance Program, Community Health Administration, Lansing, Michigan. The Florida cancer incidence data used in this report were collected by the Florida Cancer Data System (Miami, Florida) under contract with the Florida Department of Health, Tallahassee, Florida. The views expressed herein are solely those of the authors and do not necessarily reflect those of the FCDC or FDOH. Cancer incidence data from Louisiana were collected by the Louisiana Tumor Registry, Louisiana State University Health Sciences Center School of Public Health, New Orleans, Louisiana. Cancer incidence data from New Jersey were collected by the New Jersey State Cancer Registry, The Rutgers Cancer Institute of New Jersey, New Brunswick, New Jersey. Cancer incidence data from North Carolina were collected by the North Carolina Central Cancer Registry, Raleigh, North Carolina. Cancer incidence data from Pennsylvania were supplied by the Division of Health Statistics and Research, Pennsylvania Department of Health, Harrisburg, Pennsylvania. The Pennsylvania Department of Health specifically disclaims responsibility for any analyses, interpretations or conclusions. Cancer incidence data from Arizona were collected by the Arizona Cancer Registry, Division of Public Health Services, Arizona Department of Health Services, Phoenix, Arizona. Cancer incidence data from Texas were collected by the Texas Cancer Registry, Cancer Epidemiology and Surveillance Branch, Texas Department of State Health Services, Austin, Texas. Cancer incidence data from Nevada were collected by the Nevada Central Cancer Registry, Division of Public and Behavioral Health, State of Nevada Department of Health and Human Services, Carson City, Nevada.
References
- 1. Fenelon A. Geographic divergence in mortality in the United States. Population and Development Review. 2013;39:611–634. pmid:25067863
- 2. Cosby AG, McDoom-Echebiri MM, James W, Khandekar H, Brown W, Hanna HL. Growth and Persistence of Place-Based Mortality in the United States: The Rural Mortality Penalty. American Journal of Public Health. 2019; 109:155–162. pmid:30496008
- 3.
Wilmoth JR, Boe C, Barbieri M. Geographic differences in life expectancy at age 50 in the United States compared with other high-income countries. International differences in mortality at older ages: Dimensions and sources. 2011;:333–66.
- 4. Hamad R, Rehkopf DH, Kuan KY, Cullen MR. Predicting later life health status and mortality using state-level socioeconomic characteristics in early life. SSM—Population Health. 2016;2:269–76. pmid:27713921
- 5. Datta G.D., Glymour M. M., Kosheleva A., & Chen J. T. Prostate cancer mortality and birth or adult residence in the southern United States. Cancer Causes Control. 2012;23:1039–46. pmid:22547136
- 6. Patton KK, Benjamin EJ, Kosheleva A, Curtis LH, Glymour MM. Early-life antecedents of atrial fibrillation: place of birth and atrial fibrillation-related mortality. Ann Epidemiol. 2011;21:732–8. pmid:21798760
- 7. Rehkopf DH, Eisen EA, Modrek S, Mokyr Horner E, Goldstein B, Costello S, et al. Early-Life State-of-Residence Characteristics and Later Life Hypertension, Diabetes, and Ischemic Heart Disease. Am J Public Health. 2015;105:1689–95. pmid:26066927
- 8. Lynch J, Smith GD. A life course approach to chronic disease epidemiology. Annu Rev Public Health. 2005;26:1–35. pmid:15760279
- 9. Gustafsson P, Hammarström A, San Sebastián M. Cumulative contextual and individual disadvantages over the life course and adult functional somatic symptoms in Sweden. Eur J Public Health. 2015. pmid:25527526
- 10.
Montez JK, Hayward MD. Early Life Conditions and Later Life Mortality. In: Rogers RG, Crimmins EM, editors. International Handbook of Adult Mortality. Dordrecht: Springer Netherlands; 2011. p. 187–206. https://doi.org/10.1007/978-90-481-9996-9_9
- 11. Pollitt RA, Rose KM, Kaufman JS. Evaluating the evidence for models of life course socioeconomic factors and cardiovascular outcomes: a systematic review. BMC Public Health. 2005;5:7. pmid:15661071
- 12. Kahn JR, Pearlin LI. Financial strain over the life course and health among older adults. J Health Soc Behav. 2006;47:17–31. pmid:16583773
- 13. Molloy R, Smith CL, Wozniak A. Internal Migration in the United States. J Econ Perspect. 2011;25:173–96.
- 14. Phelan JC, Link BG, Tehranifar P. Social Conditions as Fundamental Causes of Health Inequalities: Theory, Evidence, and Policy Implications. J Health Soc Behav. 2010;51:S28–40. pmid:20943581
- 15. Montez JK, Hayward MD, Wolf DA. Do U.S. states’ socioeconomic and policy contexts shape adult disability? Soc Sci Med. 2017;178:115–26. pmid:28219027
- 16. Montez JK, Zajacova A, Hayward MD. Explaining Inequalities in Women’s Mortality between U.S. States. SSM Popul Health. 2016;2:561–71. pmid:27722192
- 17. Brown CC, Moore JE, Felix HC, Stewart MK, Bird TM, Lowery CL, et al. Association of State Medicaid Expansion Status With Low Birth Weight and Preterm Birth. JAMA. 2019;321:1598–609. pmid:31012935
- 18. Komro KA, Livingston MD, Markowitz S, Wagenaar AC. The Effect of an Increased Minimum Wage on Infant Mortality and Birth Weight. Am J Public Health. 2016;106:1514–6. pmid:27310355
- 19. Dhamija G. The Joint Determination of Later Life Health and Educational Outcomes by Early Life Conditions: A Multinomial Logit Analysis. Review of Market Integration. 2018;10:197–227.
- 20. Victora CG, Adair L, Fall C, Hallal PC, Martorell R, Richter L, et al. Maternal and child undernutrition: consequences for adult health and human capital. Lancet. 2008;371:340–57. pmid:18206223
- 21. Bleakley H. Malaria Eradication in the Americas: A Retrospective Analysis of Childhood Exposure. Am Econ J Appl Econ. 2010;2:1–45. pmid:24179596
- 22. Jurges H. Collateral damage: the German food crisis, educational. Journal of Health Economics. 2013;32:286–303. pmid:23237792
- 23. Qiu C, Kivipelto M, von Strauss E. Epidemiology of Alzheimer’s disease: occurrence, determinants, and strategies toward intervention. Dialogues Clin Neurosci. 2009;11:111–28. https://www.ncbi.nlm.nih.gov/pubmed/19585947.
- 24. Rhodius-Meester HFM, Liedes H, Koene T, Lemstra AW, Teunissen CE, Barkhof F, et al. Disease-related determinants are associated with mortality in dementia due to Alzheimer’s disease. Alzheimers Res Ther. 2018;10:23. pmid:29458426
- 25. Vega IE, Cabrera LY, Wygant CM, Velez-Ortiz D, Counts SE. Alzheimer’s Disease in the Latino Community: Intersection of Genetics and Social Determinants of Health. J Alzheimers Dis. 2017;58:979–92. pmid:28527211
- 26. McKhann G, Knopman D, Chertkow H, Hyman B, Kawas C, Klunk W, et al. The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging and the Alzheimer’s Association workgroup. 2011. https://www.semanticscholar.org/paper/f07eebcc0cf7d28e3b177f49eb5fefaab3cd4f98.
- 27. Mehta KM, Yeo GW. Systematic review of dementia prevalence and incidence in United States race/ethnic populations. Alzheimers Dement. 2017;13:72–83. pmid:27599209
- 28. Rabins PV, Schwartz S, Black BS, Corcoran C, Fauth E, Mielke M, et al. Predictors of progression to severe Alzheimer’s disease in an incidence sample. Alzheimers Dement. 2013;9:204–7. pmid:23123228
- 29. Spalletta G, Long JD, Robinson RG, Trequattrini A, Pizzoli S, Caltagirone C, et al. Longitudinal Neuropsychiatric Predictors of Death in Alzheimer’s Disease. J Alzheimers Dis. 2015;48:627–36. pmid:26402103
- 30. Taylor CA. Deaths from Alzheimer’s Disease—United States, 1999–2014. MMWR Morb Mortal Wkly Rep. 2017;66. pmid:28542120
- 31. Sando SB, Melquist S, Cannon A, Hutton M, Sletvold O, Saltvedt I, et al. Risk-reducing effect of education in Alzheimer’s disease. Int J Geriatr Psychiatry. 2008;23:1156–62. pmid:18484674
- 32. Zhang Y, Zhuang P, He W, Chen JN, Wang WQ, Freedman ND, et al. Association of fish and long-chain omega-3 fatty acids intakes with total and cause-specific mortality: prospective analysis of 421 309 individuals. J Intern Med. 2018;284:399–417. pmid:30019399
- 33. Burke SL, Cadet T, Alcide A, O’Driscoll J, Maramaldi P. Psychosocial risk factors and Alzheimer’s disease: the associative effect of depression, sleep disturbance, and anxiety. Aging Ment Health. 2018;22:1577–84. pmid:29077487
- 34. Dimakakou E, Johnston H, Streftaris G, Cherrie J. Exposure to Environmental and Occupational Particulate Air Pollution as a Potential Contributor to Neurodegeneration and Diabetes: A Systematic Review of Epidemiological Research. Int J Environ Res Public Health. 2018. pmid:30096929
- 35. Tsai T-L, Lin Y-T, Hwang B-F, Nakayama SF, Tsai C-H, Sun X-L, et al. Fine particulate matter is a potential determinant of Alzheimer’s disease: A systemic review and meta-analysis. Environ Res. 2019;177:108638. pmid:31421449
- 36. Schatzkin A, Subar A. F., Thompson F. E., Harlan L. C., Tangrea J., Hollenbeck A. R. Design and serendipity in establishing a large cohort. American Journal of Epidemiology. 2001;154:1119–25. pmid:11744517
- 37. Sonnega A, Faul JD, Ofstedal M, Langa KM, Phillips JWR, Weir DR. Cohort profile: the Health and Retirement Study (HRS). International Journal of Epidemiology. 2014;43:576–85. pmid:24671021
- 38. Etemadi A, Sinha R, Ward MH, Graubard BI, Inoue-Choi M, Dawsey SM, et al. Mortality from different causes associated with meat, heme iron, nitrates, and nitrites in the NIH-AARP Diet and Health Study: population based cohort study. BMJ. 2017;357. pmid:28487287
- 39. Fletcher JM. New evidence of the effects of education on health in the US: compulsory schooling laws revisited. Soc Sci Med. 2015;127:101–7. pmid:25311783
- 40. Romero JP, Benito-León J, Mitchel AJ, Trincado R, Bermejo-Pareja F. Under Reporting of Dementia Deaths on Death Certificiates using Data from A Population-Based Study (NEDICES). J Alzheimers Dis. 2014;39:741–748. pmid:24254704
- 41. Romero JP, Benito-León J, Louis ED, Bermejo-Pareja F. Under Reporting of Dementia Deaths on Death Certificates: A Systematic Review of Population-Based Cohort Studies. J Alzheimers Dis. 2014;41:213–221. pmid:24583403