Skip to main content
The Journals of Gerontology Series A: Biological Sciences and Medical Sciences logoLink to The Journals of Gerontology Series A: Biological Sciences and Medical Sciences
. 2017 Aug 9;73(4):477–483. doi: 10.1093/gerona/glx154

Genetic Risks for Chronic Conditions: Implications for Long-term Wellbeing

George L Wehby 1,2,3,4,, Benjamin W Domingue 5, Fredric D Wolinsky 1
Editor: Anne Newman
PMCID: PMC5861924  PMID: 28958056

Abstract

Background

Relationships between genetic risks for chronic diseases and long-run wellbeing are largely unexplored. We examined the associations between genetic predispositions to several chronic conditions and long-term functional health and socioeconomic status (SES).

Methods

We used data on a nationally representative sample of 9,317 adults aged 65 years or older from the 1992 to 2012 Health and Retirement Survey (HRS) in the US. Survey data were linked to genetic data on nearly 2 million single-nucleotide polymorphisms (SNPs). We measured individual-level genetic predispositions for coronary-artery disease, type 2 diabetes (T2D), obesity, rheumatoid arthritis (RA), Alzheimer’s disease, and major depressive disorder (MDD) by polygenic risk scores (PRS) derived from genome-wide association studies (GWAS). The outcomes were self-rated health, depressive symptoms, cognitive ability, activities of everyday life, educational attainment, and wealth. We employed regression analyses for the outcomes including all polygenic scores and adjusting for gender, birth period, and genetic ancestry.

Results

The polygenic scores had important associations with functional health and SES. An increase in genetic risk for all conditions except T2D was significantly (p < .01) associated with reduced functional health and socioeconomic outcomes. The magnitudes of functional health declines were meaningful and in many cases equivalent in magnitude to several years of aging. These associations were robust to several sensitivity checks for ancestry and adjustment for parental educational attainment and age at death or the last interview if alive.

Conclusion

Stronger genetic predispositions for leading chronic conditions are related to worse long-run health and SES outcomes, likely reflecting the adverse effects of the onset of these conditions on one’s wellbeing.

Keywords: Chronic conditions, Polygenic scores, Wellbeing


A majority of older adults (>60%) live with two or more chronic conditions (1). Twin studies have long established that genes are contributors to chronic conditions such as cardiovascular disease, diabetes, obesity, RA, Alzheimer’s disease (AD), and depression (2–10). Only recently, however, has it become possible to measure individual-level risks for these chronic diseases using molecular genetic data from GWAS. These studies facilitate individual-level quantification of genetic risks for specific conditions using PRS based on a person’s genotype and the GWAS-derived genetic effects on the conditions (11). GWAS-derived PRS have shown promising results in explaining variation in individual outcomes in several settings (12–16).

Although evidence is accumulating about the processes through which genetic predispositions impact the onset of chronic diseases, less is known about the longer term implications. A natural hypothesis is that genetic risks for major chronic conditions are related to worse long-term functional health outcomes (eg, self-reported health status, depressive symptoms, cognitive function, or limitations in everyday activities). Such hypothesis has, of yet, little to no empirical support. Genetic predisposition can influence these functional health outcomes through the onset, severity, and prognosis of chronic conditions including preclinical effects. All else equal, stronger genetic risks likely worsen functional health outcomes. Furthermore, even though moderate to strong genetic influence on several functional health outcomes among older adults has been documented in twin studies, empirical evidence on specific genetic pathways is limited (17). Recent evidence indicates that greater genetic predisposition to obesity measured by GWAS-derived PRS is associated with higher medicare expenditures among older adults (18), providing support for the hypothesis that measurable individual-level genetic risks for chronic conditions can have long-run effects on health and even health services use and expenditures.

By impacting functional health, genetic predisposition to chronic conditions may also reduce SES attainment. Recent work has employed two approaches to evaluate how genes may influence such attainments. One has been to identify genetic variants associated with specific SES outcomes (eg, education (19), social deprivation, and household income (20)). A second approach has been to use PRS as instruments for anthropometric traits like height and body weight in an attempt to identify their effects on SES (21). There is, however, little empirical work that examines the effects of genetic predispositions to chronic conditions on SES attainment, especially on long-run SES outcomes among older adults like total wealth. Examining these links is important for understanding how heritable biological risks, in the form of measurable genetic predispositions to chronic diseases, are related to long-term wellbeing and successful aging (22,23).

The literature to date has focused on the effects of self-reported chronic conditions on long-term health and SES outcomes (24,25). However, reporting errors (26,27), undiagnosed conditions, and confounders resulting from complex genetic and environmental etiologies are key challenges. Therefore, focusing on genetic predisposition to onset of chronic disease offers important advantages over self-reported conditions.

We begin to fill in these knowledge gaps by examining the associations between genetic predispositions to major chronic conditions and long-run functional health and SES attainment. Specifically, we examined how PRS for coronary artery disease (CAD) (28), T2D (29), body mass index (BMI) (30), RA (31), AD (32), and MDD (33) derived from meta-GWAS are related to several health and SES outcomes among older adults. Our goal is not to estimate causal effects of these chronic conditions on these outcomes by using the PRS as instrumental variables (34), which requires strong assumptions that are untenable in the context of the genome-wide PRS that we deploy. Instead, our aim is to understand whether these PRS are associated with a broader range of outcomes associated with long-term wellbeing.

The premise of our study is that the chronic conditions we investigate are heavily influenced by genetic risks which can be measured at the individual level using genome-wide PRS. As noted above, there is strong evidence for genetic influences on these conditions through twin studies and more recently through GWAS. We also show below evidence that the PRS are significantly predictive of the incidence of these chronic conditions. The onset and severity of these chronic conditions can substantially affect functional health such as general health status, cognition, psychosocial health, and mobility. By increasing an individual’s predisposition to the onset and possibly the severity of chronic conditions, genetic risks captured by the PRS can exert lifelong and cumulative adverse effects on functional health outcomes. Therefore, we hypothesize that PRS capturing increased risks for CAD, T2D, obesity, RA, AD, and MDD are associated with declines in functional health both in physical and mental domains. No prior study has directly tested this hypothesis. As noted above, genetic risks captured through PRS are not influenced by report errors and biases and are less prone to confounding compared with self-reports of chronic conditions. They can also capture effects from preclinical manifestations.

Our second objective was to examine whether the individual-level genetic risks for such chronic conditions are related to worse SES, including educational attainment and wealth. As noted above, early onset and preclinical manifestations of these chronic conditions can have negative influences on physical, cognitive, and emotional health that accumulate over time, which may in turn reduce academic achievement. Furthermore, the declines in functional health and educational attainment associated with the onset and severity of these conditions can also reduce employment and economic productivity, which would in turn decrease wealth accumulation. Therefore, our second hypothesis is that greater genetic predisposition to the above-mentioned chronic conditions is associated with lower educational attainment and wealth. This is the first study to directly test and evaluate the effects of these specific genetic risks on education and wealth.

Design and Methods

Data

Our data are from the HRS which provides biennial, longitudinal data beginning in 1992 for a nationally representative sample of individuals 50 years or older and their spouses. Baseline interviews were followed by biennial reinterviews currently available through 2012. The HRS obtained DNA samples from participants who were alive in 2006 and 2008. These DNA samples were genotyped in a GWAS panel of nearly 2 million SNPs. We linked the HRS GWAS data to survey data from 1992 to 2012.

Of the 12,507 genotyped participants, 12,358 passed quality control filters. Because allele frequencies may differ by ancestry resulting in population stratification bias (35), we focused on the 9,453 self-reported non-Hispanic whites with GWAS data. We also show results using all respondents after adjusting for ancestry. We focused on life-long effects among individuals aged 65 or older at their last HRS interview since they have greater variation in the key well-being outcomes we study like poor/fair health, activities of daily living (ADLs), or cognition (discussed below) than younger adults and a longer period for genetic effects to manifest. Furthermore, only 2.3% of the sample has their last HRS interview before age 65. Our main analytical sample includes 7,338 self-reported non-Hispanic whites born before 1948. When including non-whites, the sample includes 9,317 individuals.

Outcomes

The health outcomes were self-reported health, depressive symptoms, cognitive ability, ADLs, and instrumental ADLs (IADLs). These outcomes were based on the responses from the last survey of each participant. We evaluated self-reported health using the traditional five-category ranking scale and as a binary outcome contrasting fair or poor health with good, very good, or excellent health. Limitations in everyday activities were based on the number of reported ADLs or IADLs that could not be performed or were performed with difficulty. The ADLs included bathing, dressing, eating, getting in and out of bed, and walking across a room, and the IADLs included using the phone, managing money, taking medications, shopping for groceries, and preparing hot meals. We evaluated depressive symptoms using the Center for Epidemiologic Studies Depression 8-item scale, both as a count of the symptoms endorsed and a binary outcome for being at or above the threshold for depression risk (three or more). Cognitive ability was measured by the Telephone Interview for Cognitive Status (36), which sums the number of correct items from a seven item mental status test, ten item immediate and delayed word recall tests, and serial sevens.

We examined SES via education and total wealth. We measured education via total years of formal schooling and by two binary indicators (less than a high school education and completion of college). Total wealth included all self-reported assets minus debts from the last survey.

Polygenic Risk Scores (PRS)

We calculated genome-wide PRS for BMI, CAD, T2D, MDD, AD, and RA. First, we evaluated the quality of the genetic data by applying standard filters for minor allele frequencies, minimum genotyping success, and Hardy-Weinberg equilibrium (leaving ~1.7M SNPs). For each PRS, we used the most recently available meta-GWAS (note that none included data from the HRS) (29,30,32,33,37,38). A weighted mean for each individual across all SNPs that were available in both the meta-GWAS and the HRS-GWAS was then calculated using the SNP effects in the meta-GWAS as weights. SNPs that are stronger predictors of chronic disease are weighted more heavily, which is the standard approach for deriving PRS (39–41). Increases in each of the chronic diseases, PRS, indicate a higher genetic risk. The PRS were then standardized on the HRS-GWAS sample in order to estimate the effects of a one standard deviation (one-SD) change compared with the sample mean.

Statistical Analysis

We begin by showing that the PRS were significantly related to their respective chronic diseases based on participant self-reports of ever having been told by a health professional that they had these conditions using logistic regression. We can do this for all conditions except for RA since the survey does not differentiate between RA and other arthritis types. Next, we examined the simultaneous associations of the PRS in regressions for each of the health and SES outcomes by including all PRS jointly in each model using ordinary least squares for continuous survey outcomes (except wealth, for which we used median regression due to its skewness) and logistic regressions for binary outcomes. All regressions adjusted for gender, birth period (1916, 1916–1920, 1921–1925, 1926–1930, 1931–1935, and 1941–1947 with 1936–1940 as the reference group), and the first ten principal components for ancestry were estimated from the HRS-GWAS data. Because our goal was to examine the total effects of the PRS on the functional and the SES outcomes through all potential mechanisms, our main models did not adjust for mediating pathways as that would result in partial effects of PRS. For instance, we did not adjust for self-reported chronic conditions or more smoking or exercise as these could be pathways through which genetic predisposition affects the functional health and SES. In additional models, however, we added the self-reported chronic conditions to evaluate if genetic predisposition captured additional information, such as preclinical manifestation, illness severity, and measurement error in self-reports of conditions.

Results

The sample summary statistics for the main analytical sample of non-Hispanic whites are shown in Table 1. The prevalence of reported chronic conditions ranged from 8% (memory problems) to 39% (heart problems). Nearly, one-third reported being in poor or fair health and 20% were above the Center for Epidemiologic Studies Depression threshold indicating depression risk. One-fifth did not graduate high school, and nearly one-quarter were college graduates. Median net wealth was about $60,000.

Table 1.

Descriptive Statistics for Self-reported Non-Hispanic White Health and Retirement Study Participants 65 Years Old or Older

Health conditions
 Obesity 24.7%
 Heart problems 39.2%
 Diabetes 23.5%
 Psychiatric problems 18.6%
 Memory problems 8.3%
Functional health outcomes
 Self-reported health 2.96 (1.10)
 Poor or fair self-reported health 30.4%
 Depressive symptoms 1.36 (1.85)
 Depression threshold 20.3%
 Cognition summary score 21.65 (5.16)
 ADL difficulties 0.51 (1.15)
 IADL difficulties 0.25 (0.68)
Socioeconomic Status
 Years of education 13.0 (2.6)
 Less than a high school education 19.6%
 College graduate 23.0%
 Total wealth (in $10,000 increments) 58.5 (120.1)
Female (vs Male) 57.0%
Birth year
 ≤1915 1.7%
 1916–1920 5.3%
 1921–1925 10.6%
 1926–1930 15.8%
 1931–1935 20.4%
 1936–1940 23.9%
 1941–1947 22.4%

Note: The table shows means and SDs in parentheses for continuous variables and rates for categorical variables. The outcomes are measured from each participant’s last interview with non-missing data (survey data available between 1992 and 2012) for a total sample of 7,338 individuals (7,324 individuals for cognition summary score and 7,336 for years of education). Obesity is based on the last BMI exceeding 30. Heart problems, diabetes, psychiatric problems, and memory problems are based on the report being told by a health care professional that the participant has had these conditions. Self-reported health was coded 1 for excellent, 2 for very good, 3 for good, 4 for fair, and 5 for poor. Fair or poor self-rated health is a binary contrast indicator. Depressive symptoms are measured by a number of eight endorsed symptoms. Depression threshold reflects 3 or more depressive symptoms. Cognition is the summary scale score ranging from 0 (worst) to 35 (best) reflecting immediate and delayed word recall, mental status, and serial 7s derived from the Telephone Interview for Cognitive Status. ADLs and IADLs are the number of difficulties with activities of daily living or instrumental activities of daily living. ADL = activities of daily living, IADL = instrumental activities of daily living.

The PRS were significantly related to their respective health conditions captured in the survey (Supplementary Table S1). All regression results are reported in terms of a one-SD change in the PRS. Some of the associations were large, with more than doubled odds for obesity (p < .001) and diabetes (p < .001). Other associations were more moderate, predicting increased odds by 55% for memory problems (p < .05), by 21% for heart diseases (p < .001), and 13% for psychiatric problems (p < .01). The differences in these associations are due to both differences in heritability of the different conditions and variability in the precision of GWAS estimates (eg, as a function of GWAS sample size). Also, self-reported error varies between these conditions in the HRS dataset and is markedly less for certain conditions such as diabetes than others such as heart problems (26).

The associations of the PRS with the functional health outcomes were considerable (Table 2). The PRS for BMI, CAD, MDD, AD, and AD were significantly related to a variety of worse outcomes. In contrast, the T2D PRS was not associated with worse functional health outcomes, and there was no evidence that this overall null finding resulted from multicollinearity with the other PRS (the variation-inflation factor for the T2D score was 1.3, well below the threshold of 10 indicating a potential collinearity issue). A sensitivity analysis that included only the T2D PRS also failed to find any associations with most of the functional health outcomes (Supplementary Table S2). When including non-whites, similar results were largely observed although results for the MDD score were attenuated (Supplementary Table S3).

Table 2.

Associations of One-SD Increases in PRS with Functional Health Outcomes among Self-reported Non-Hispanic White Health and Retirement Study Participants 65 Years Old or Older

Polygenic risk scores Self-reported health,
β [95% CI]
Poor or fair self- reported health, OR [95% CI] Depressive symptoms, β [95% CI] Depression threshold, OR [95% CI] Cognition summary score, β [95% CI] ADL difficulties, β [95% CI] IADL difficulties, β [95% CI]
BMI 0.15*** 1.28*** 0.29*** 1.52*** −0.16 0.12*** 0.00
[0.09, 0.21] [1.14, 1.44] [0.19, 0.39] [1.33, 1.74] [−0.41, 0.09] [0.07, 0.18] [−0.03, 0.04]
Coronary artery disease 0.06*** 1.10*** 0.04 1.06 −0.18*** 0.01 0.01
[0.02, 0.09] [1.03, 1.17] [−0.02, 0.09] [0.98, 1.14] [−0.32, −0.05] [−0.02, 0.05] [−0.01, 0.03]
Type 2 diabetes −0.07* 0.91 −0.02 0.97 0.18 −0.05 −0.06**
[−0.15, 0.00] [0.78, 1.07] [−0.15, 0.12] [0.82, 1.16] [−0.15, 0.51] [−0.13, 0.03] [−0.10, −0.01]
Major depressive disorder 0.05** 1.09** 0.08** 1.12** −0.14* 0.03 0.02
[0.01, 0.08] [1.01, 1.17] [0.02, 0.14] [1.03, 1.22] [−0.29, 0.02] [−0.01, 0.07] [−0.01, 0.04]
Alzheimer’s 0.14*** 1.33*** 0.33*** 1.48*** −0.85*** 0.07 0.07**
[0.04, 0.24] [1.09, 1.64] [0.16, 0.49] [1.17, 1.86] [−1.27, −0.43] [−0.04, 0.17] [0.01, 0.13]
Rheumatoid arthritis 0.06*** 1.10** 0.07* 1.07 −0.23** 0.02 −0.00
[0.02, 0.11] [1.01, 1.20] [−0.01, 0.14] [0.97, 1.19] [−0.41, −0.04] [−0.03, 0.06] [−0.03, 0.03]
N 7,338 7,338 7,338 7,338 7,324 7,338 7,338

Note: The table shows the regression coefficients [β] for continuous outcomes and odds ratios [OR] for binary outcomes of a one-SD increase in the polygenic risk scores (with 95% confidence intervals in brackets) in regressions for functional health outcomes. The polygenic risk scores were all included together in each regression model. A separate regression model was estimated for each functional health outcome. Models were estimated using linear regression for continuous outcomes and logistic regression for binary outcomes. All models were adjusted for gender, birth period, and the first 10 ancestry principal components from the GWAS data. Self-reported health was coded 1 for excellent, 2 for very good, 3 for good, 4 for fair, and 5 for poor. Fair or poor self-rated health is a binary contrast indicator. Depressive symptoms are measured by a number of eight endorsed symptoms. Depression threshold reflects three or more depressive symptoms. Cognition is the summary scale score ranging from 0 (worst) to 35 (best) reflecting immediate and delayed word recall, mental status, and serial 7s derived from the Telephone Interview for Cognitive Status. ADLs and IADLs are the number of difficulties with activities of daily living or instrumental activities of daily living. All outcomes are measured at the last follow-up interview with the participant. * p < .1, ** p < .05, *** p < .01. BMI = body mass index.

Among non-Hispanic whites, the associations with self-reported health were largest for the BMI and AD PRS, with a one-SD increase in PRS corresponding to nearly 5% reduction in self-reported health relative to the sample mean (regression coefficients in Table 2 divided by sample mean of self-reported health in Table 1 and then multiplied by 100) and 30% greater odds of reporting poor or fair health. The CAD, MDD, and RA PRS had smaller yet meaningful associations, including 10% increase in odds of poor or fair health with one-SD PRS increase.

The largest associations with depressive symptoms and the clinical threshold were for the BMI and AD PRS, with 50% greater depression odds per one-SD score increase. The MDD PRS was also associated with greater depression odds but by only 12% suggesting that the BMI and AD PRS were related to a wider range of depressive states, including less severe forms of depression than the MDD PRS.

The BMI score was not significantly related to cognitive ability, but the PRS for CAD, AD, and RA were significantly related to declines in the cognitive ability score by 1%, 4%, and 1% relative to the sample means, respectively, with one-SD score increase. The MDD PRS was also marginally associated with a 1% decline in cognitive ability. Fewer associations were observed with ADLs than the other functional health outcomes. Only the BMI PRS was significantly related to more ADLs by 24% at sample mean. IADL limitations were greater with higher AD PRS by 28%. Interestingly, the T2D PRS was associated with an improvement in similar magnitude.

All PRS except for T2D were also significantly related to SES declines. One-SD increases in genetic risks for BMI, CAD, and AD were associated with 0.3 fewer educational years while the RA PRS was related to 0.2 fewer years (Table 3). The increase in odds of less than a high school education ranged from 34% with the BMI PRS to 16% for the MDD PRS, while the decline in odds of college graduation ranged from 21% for CAD to 13% for RA. The BMI, CAD, AD, and RA PRS were consistently related to lower schooling on all three educational attainment measures. In contrast, the MDD PRS was related to a 10% increase in the odds of college graduation.

Table 3.

Associations of One-SD Increases in PRS with Socioeconomic Status among Self-reported Non-Hispanic White Health and Retirement Study Participants 65 Years Old or Older

Polygenic risk scores Years of education, β [95% CI] Less than a high school education, OR [95% CI] College graduate, OR [95% CI] Total wealth (in $10,000 increments), β [95% CI]
BMI −0.27*** 1.34*** 0.83*** −5.18***
[−0.41, −0.14] [1.17, 1.54] [0.73, 0.94] [−7.48, −2.89]
Coronary artery disease −0.27*** 1.20*** 0.79*** −2.13***
[−0.35, −0.19] [1.11, 1.30] [0.73, 0.85] [−3.46, 0.79]
Type 2 diabetes 0.16* 0.86 1.11 1.40
[−0.03, 0.34] [0.72, 1.04] [0.93, 1.32] [−2.01, 4.80]
Major depressive disorder 0.00 1.16*** 1.10** −0.98
[−0.08,0.09] [1.06, 1.26] [1.01, 1.19] [−2.46, 0.51]
Alzheimer’s −0.34*** 1.26* 0.80* −4.59**
[−0.58, −0.10] [0.99,1.59] [0.64, 1.00] [−8.87, −0.32]
Rheumatoid arthritis −0.22*** 1.25*** 0.87*** −1.74*
[−0.32, −0.12] [1.13, 1.38] [0.79, 0.96] [3.49, 0.01]
N 7,336 7,337 7,337 7,338

Note: The table shows the regression coefficients [β] for continuous outcomes and odds ratios [OR] for binary outcomes of a one-SD increase in the polygenic risk scores (with 95% confidence intervals in brackets) in regressions for socioeconomic status outcomes. The polygenic risk scores were all included together in each regression model. A separate regression model was estimated for each functional health outcome. Models were estimated using linear regression for years of education, logistic regression for binary outcomes, and median regression (with 1,000 bootstrap replications for standard errors) for wealth. All models were adjusted for gender, birth period, and the first 10 ancestry principal components from the GWAS data. Total wealth is coded in $10,000 increments. All outcomes are measured at the last follow-up interview with the participant. * p < .1, ** p < .05, *** p < .01. BMI = body mass index.

Finally, the PRS for BMI and AD were significantly associated with reductions in median wealth of about $5,000, while the CAD and RA PRS were associated with nearly $2,000 lower wealth. Similar results were generally observed when adding non-whites, but with generally less pronounced associations especially for MDD (Supplementary Table S4).

Discussion

The majority of older adults have two or more chronic diseases and must cope with their attendant illness burdens through their remaining life. Understanding the relationships between the underlying mechanisms for these conditions and long-term wellbeing including genetic links may offer insights into strategies for promoting successful and healthy aging (22,23). We show that genetic predispositions to major chronic health conditions are strongly and consistently related to several aspects of functional health and SES among older adults. We assess the meaningfulness and clinical relevance of these associations by comparing them to changes in the same health outcomes with age, estimated from regressions of the outcomes on age adjusting for gender (and race/ethnicity) in the same sample (Supplementary Table S5). For most outcomes, the effects of a one-SD PRS increase are equivalent in magnitude to multiple years of aging, especially for self-reported health and depressive symptoms. For example, the decline in self-reported health associated with a one-SD increase in the BMI PRS is equivalent to nearly 7 years of aging, while the decline in cognitive ability with a one-SD increase in the AD PRS is equivalent to nearly 3 years of aging. These findings suggest that heritable biological risks for chronic conditions captured through PRS are associated with meaningful long-run declines in functional health and lower lifelong socioeconomic achievements. The key mechanism underlying these associations is likely the impact of onset of chronic conditions on the studied outcomes. Because genes are inherited at conception, the interplay between genes and these outcomes may start early in life, in the form of earlier onset of chronic conditions or preclinical presence, and intensify over time.

Although all the PRS other than for T2D have meaningful associations with both functional health and SES outcomes, the BMI and the AD PRS show overall larger effects. Obesity can have an early onset beginning in childhood and can have lifelong effects on multiple health conditions (42) as well as socioeconomic achievement including income (21). Therefore, it is not surprising that the BMI PRS has systematic and prominent associations with the examined outcomes. The predictive power of the BMI PRS may also be due to the lower measurement error in BMI and especially the large sample used for its GWAS compared with other conditions. The AD PRS had the largest association with cognitive decline among all scores by at least three times. Such cognitive declines may be key mediators for the associations between the AD PRS and reduced physical and mental health wellbeing. Furthermore, nearly 5% of AD cases may have early onset prior to 65 (43). The smaller associations with the other conditions may potentially reflect a combination of more effective medical management such as with medications as well as more delayed effects, in addition to less predictive power of the PRS.

Our study adds to the accumulating evidence that genetic effects characterized by GWAS are strongly associated with secondary outcomes related to the GWAS target traits, as shown recently in greater genetic predisposition to obesity being associated with higher medicare expenditures (18). Our findings have important implications. The functional health indicators that we studied—self-reported health, depression, cognitive health, and ADLs—are among the official yardsticks used to measure and track population health and evaluate whether public health policies are successful (44). Adjusting for underlying genetic predispositions enables a more accurate assessment of the role of genetic versus non-genetic factors in the etiological process leading to these functional health outcomes. Further, these genetic risks may have implications for modifying health services use and expenditures (18). Evaluating these effects may ultimately be useful for informing the designing of more effective care management programs and treatment plans that ensure adequate access to high quality and effective health services as well as access to prevention and health promotion programs early in life that may reduce the long-term adverse effects of genetic risks.

Since parents bequeath their children more than just genes, including early investments in their development, SES, and sometimes wealth transfers, one should consider the possibility that the genetic associations we observed partly reflect intergenerational effects through parents’ genetic risks and health. Observed associations may be due partly to the conditions in the previous generation since parental genetic predispositions for chronic conditions affect their own health, which in turn may affect their children’s SES and health (eg, a chronic condition limits the parent’s economic opportunities). Although this does not bias the magnitude of the overall genetic risk that we estimated, without access to parental PRS, the contributions of the individual and parental genetic risks cannot be disentangled. The HRS, however, did obtain data on parental educational attainment, and we used these data to further adjust the models for the functional health and SES outcomes to approximate how much of the polygenic risk score associations might reflect parental effects. For both functional health (Supplementary Table S6) and SES (Supplementary Table S7), comparable results were observed for most of the PRS. The exception was for the Alzheimer’s PRS, which had smaller and mostly insignificant associations with SES attainment after adjusting for parental education, although its strong associations with the functional health outcomes remained. HRS also included data on whether the participants’ parents were alive at each participant’s interview and their age at death or the last interview. After re-estimating the models with further adjustments for these data, we again observed comparable results for most of the PRS (Supplementary Tables S8 and S9). Taken together, these additional analyses suggest that the associations we found were not confounded by these indicators of parental SES and health. Nonetheless, examining this question in future research with intergenerational data on genetic risks, and long-term health and SES outcomes would be useful to separate parental and individual effects.

Because of errors in self-reported chronic conditions and since the PRS may capture early onset, preclinical manifestation, and illness severity, the associations we found are not fully captured by the self-reported conditions themselves. Indeed, adjusting for the self-reported health conditions from the survey data that are closest to the conditions corresponding to the PRS attenuated but did not eliminate most of the associations (Supplementary Tables S10 and S11). This shows that the PRS capture additional information about predisposition to chronic conditions above and beyond self-reported data and demonstrate the value of directly evaluating the total effects of genetic risks on functional health and SES attainment and that self-reported conditions are not optimal measures of genetic risks.

Two caveats of our study are worth noting. The PRS include measurement error which would attenuate associations towards the null. Larger meta-GWAS in the future will enable the construction of more powerful and predictive PRS. Another potential issue is mortality of HRS participants prior to DNA collection in 2006 and 2008. If individuals with greater genetic risks for chronic disease and worse functional health outcomes were more likely to die, the PRS associations that we estimated on the sample with genetic data could also be attenuated as suggested in other work (45). Both of these issues would suggest that the estimates of PRS associations with the functional health and SES outcomes that we obtain are lower bound and that such associations would even be larger in the population. Finally, it is worth highlighting that our conclusions are not dependent on multiple testing as our goal was not to test for separate genetic predictors of independent outcomes but instead to jointly evaluate how genetic predispositions to several chronic conditions are associated with well-being indicators. However, most of the PRS associations remain statistically significant even when using a strict approach to account for multiple comparisons like Bonferroni correction. For example, four of the five PRS associations with self-reported health remain statistically significant using a Bonferroni adjustment for evaluating seven functional health outcomes.

Supplementary Material

Supplementary data is available at The Journals of Gerontology Series A: Biological Sciences and Medical Sciences online.

Supplementary Material

Supplementary Tables S1-S11

Acknowledgments

Our study uses genetic data from the Health and Retirement which is sponsored by the National Institute on Aging (grant numbers U01AG009740, RC2AG036495, and RC4AG039029) and was conducted by the University of Michigan. The genetic data was obtained through dbGaP.

References

  • 1. Ward BW, Schiller JS. Prevalence of multiple chronic conditions among US adults: estimates from the National Health Interview Survey, 2010. Prev Chronic Dis. 2013;10:E65. doi: 10.5888/pcd10.120203 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Carlsson S, Ahlbom A, Lichtenstein P, Andersson T. Shared genetic influence of BMI, physical activity and type 2 diabetes: a twin study. Diabetologia. 2013;56:1031–1035. doi: 10.1007/s00125-013-2859-3 [DOI] [PubMed] [Google Scholar]
  • 3. Kendler KS, Gatz M, Gardner CO, Pedersen NL. A Swedish national twin study of lifetime major depression. Am J Psychiatry. 2006;163:109–114. doi: 10.1176/appi.ajp.163.1.109 [DOI] [PubMed] [Google Scholar]
  • 4. Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet. 1997;27:325–351. [DOI] [PubMed] [Google Scholar]
  • 5. Silventoinen K, Rokholm B, Kaprio J, Sørensen TI. The genetic and environmental influences on childhood obesity: a systematic review of twin and adoption studies. Int J Obes (Lond). 2010;34:29–40. doi: 10.1038/ijo.2009.177 [DOI] [PubMed] [Google Scholar]
  • 6. Gatz M, Reynolds CA, Fratiglioni L et al. Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry. 2006;63:168–174. doi: 10.1001/archpsyc.63.2.168 [DOI] [PubMed] [Google Scholar]
  • 7. MacGregor AJ, Snieder H, Rigby AS et al. Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins. Arthritis Rheum. 2000;43:30–37. doi: 10.1002/1529-0131(200001)43: 1<30::AID-ANR5>3.0.CO;2-B [DOI] [PubMed] [Google Scholar]
  • 8. Zdravkovic S, Wienke A, Pedersen NL, Marenberg ME, Yashin AI, De Faire U. Heritability of death from coronary heart disease: a 36-year follow-up of 20 966 Swedish twins. J Intern Med. 2002;252:247–254. [DOI] [PubMed] [Google Scholar]
  • 9. Evans A, Van Baal GC, McCarron P et al. The genetics of coronary heart disease: the contribution of twin studies. Twin Res. 2003;6:432–441. doi: 10.1375/136905203770326439 [DOI] [PubMed] [Google Scholar]
  • 10. Wienke A, Herskind AM, Christensen K, Skytthe A, Yashin AI. The heritability of CHD mortality in Danish twins after controlling for smoking and BMI. Twin Res Hum Genet. 2005;8:53–59. doi: 10.1375/1832427053435328 [DOI] [PubMed] [Google Scholar]
  • 11. Plomin R, Haworth CM, Davis OS. Common disorders are quantitative traits. Nat Rev Genet. 2009;10:872–878. doi: 10.1038/nrg2670 [DOI] [PubMed] [Google Scholar]
  • 12. Belsky DW, Moffitt TE, Houts R et al. Polygenic risk, rapid childhood growth, and the development of obesity: evidence from a 4-decade longitudinal study. Arch Pediatr Adolesc Med. 2012;166:515–521. doi: 10.1001/archpediatrics.2012.131 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Domingue BW, Belsky DW, Harris KM, Smolen A, McQueen MB, Boardman JD. Polygenic risk predicts obesity in both white and black young adults. PLoS One. 2014;9:e101596. doi: 10.1371/journal.pone.0101596 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Purcell SM, Wray NR, Stone JL, Visscher PM, O’Donovan MC, Sullivan PF et al. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature. 2009;460:748–752. doi: 10.1038/nature08185 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Wray NR, Goddard ME, Visscher PM. Prediction of individual genetic risk to disease from genome-wide association studies. Genome Res. 2007;17:1520–1528. doi: 10.1101/gr.6665407 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Krapohl E, Euesden J, Zabaneh D, Pingault JB, Rimfeld K, von Stumm S et al. Phenome-wide analysis of genome-wide polygenic scores. Mol Psychiatry. 2016;21:1188–1193. doi: 10.1038/mp.2015.126 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Foebel AD, Pedersen NL. Genetic influences on functional capacities in aging. Gerontologist. 2016;56 (Suppl 2):S218–S229. doi: 10.1093/geront/gnw006 [DOI] [PubMed] [Google Scholar]
  • 18. Wehby GL, Domingue B, Ullrich F, Wolinsky F. Genetic predisposition to obesity and medicare expenditures. J Gerontol A Biol Sci Med Sci 2017. doi: 10.1093/gerona/glx062 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Okbay A, Beauchamp JP, Fontana MA et al. ; LifeLines Cohort Study Genome-wide association study identifies 74 loci associated with educational attainment. Nature. 2016;533:539–542. doi: 10.1038/nature17671 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Hill WD, Hagenaars SP, Marioni RE et al. Molecular genetic contributions to social deprivation and household income in UK Biobank. Curr Biol. 2016;26:3083–3089. doi: 10.1016/j.cub.2016.09.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Tyrrell J, Jones SE, Beaumont R et al. Height, body mass index, and socioeconomic status: mendelian randomisation study in UK Biobank. BMJ. 2016;352:i582. doi: https://doi.org/10.1136/bmj.i582 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Institute of Medicine [IOM]. Living well with chronic illness: a call for public health action. Washington, DC: The National Academies Press; 2012. doi: https://doi.org/10.17226/13272 [Google Scholar]
  • 23. Rowe JW, Kahn RL.. Successful aging: the MacArthur foundation study. New York: Pantheon; 1998. [Google Scholar]
  • 24. Case A, Fertig A, Paxson C. The lasting impact of childhood health and circumstance. J Health Econ. 2005;24:365–389. doi: 10.1016/j.jhealeco.2004.09.008 [DOI] [PubMed] [Google Scholar]
  • 25. Currie J, Stabile M, Manivong P, Roos LL. Child health and young adult outcomes. Journal of Human Resources. 2010;45:517–548. doi: 10.3386/w14482 [Google Scholar]
  • 26. Wolinsky FD, Jones MP, Ullrich F, Lou Y, Wehby GL. The concordance of survey reports and medicare claims in a nationally representative longitudinal cohort of older adults. Med Care. 2014;52:462–468. doi: 10.1097/MLR.0000000000000120 [DOI] [PubMed] [Google Scholar]
  • 27. Wolinsky FD, Jones MP, Ullrich F, Lou Y, Wehby GL. Cognitive function and the concordance between survey reports and medicare claims in a nationally representative cohort of older adults. Med Care. 2015;53:455–462. doi: 10.1097/MLR.0000000000000338 [DOI] [PubMed] [Google Scholar]
  • 28. Nikpay M, Goel A, Won HH et al. A comprehensive 1,000 genomes-based genome-wide association meta-analysis of coronary artery disease. Nat Genet. 2015;47:1121–1130. doi: 10.1038/ng.3396 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Morris AP, Voight BF, Teslovich TM et al. ; Wellcome Trust Case Control Consortium; Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC) Investigators; Genetic Investigation of ANthropometric Traits (GIANT) Consortium; Asian Genetic Epidemiology Network–Type 2 Diabetes (AGEN-T2D) Consortium; South Asian Type 2 Diabetes (SAT2D) Consortium; DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium Large-scale association analysis provides insights into the genetic architecture and pathophysiology of type 2 diabetes. Nat Genet. 2012;44:981–990. doi: 10.1038/ng.2383. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Locke AE, Kahali B, Berndt SI et al. ; LifeLines Cohort Study; ADIPOGen Consortium; AGEN-BMI Working Group; CARDIOGRAMplusC4D Consortium; CKDGen Consortium; GLGC; ICBP; MAGIC Investigators; MuTHER Consortium; MIGen Consortium; PAGE Consortium; ReproGen Consortium; GENIE Consortium; International Endogene Consortium Genetic studies of body mass index yield new insights for obesity biology. Nature. 2015;518:197–206. doi: 10.1038/nature14177 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Okada Y, Wu D, Trynka G et al. ; RACI consortium; GARNET consortium Genetics of rheumatoid arthritis contributes to biology and drug discovery. Nature. 2014;506:376–381. doi: 10.1038/nature12873 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Lambert JC, Ibrahim-Verbaas CA, Harold D et al. ; European Alzheimer’s Disease Initiative (EADI); Genetic and Environmental Risk in Alzheimer’s Disease; Alzheimer’s Disease Genetic Consortium; Cohorts for Heart and Aging Research in Genomic Epidemiology Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet. 2013;45:1452–1458. doi: 10.1038/ng.2802 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Ripke S, Wray NR, Lewis CM, Hamilton SP, Weissman MM, Breen G et al. A mega-analysis of genome-wide association studies for major depressive disorder. Mol Psychiatry. 2013;18:497–511. doi: 10.1038/mp.2012.21 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Wehby GL, Ohsfeldt RL, Murray JC. ‘Mendelian randomization’ equals instrumental variable analysis with genetic instruments. Stat Med. 2008;27:2745–2749. doi: 10.1002/sim.3255 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Cardon LR, Palmer LJ. Population stratification and spurious allelic association. Lancet. 2003;361:598–604. doi: 10.1016/S0140-6736(03)12520-2 [DOI] [PubMed] [Google Scholar]
  • 36. Brandt J., Spencer M., M. F. The telephone interview for cognitive status. Neuropsychiatry Neuropsychol Behav Neurol. 1988:111–117. [Google Scholar]
  • 37. Schunkert H, König IR, Kathiresan S et al. ; Cardiogenics; CARDIoGRAM Consortium Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Nat Genet. 2011;43:333–338. doi: 10.1038/ng.784 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Stahl EA, Raychaudhuri S, Remmers EF et al. ; BIRAC Consortium; YEAR Consortium Genome-wide association study meta-analysis identifies seven new rheumatoid arthritis risk loci. Nat Genet. 2010;42:508–514. doi: 10.1038/ng.582 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Domingue BW, Belsky D, Conley D, Harris KM, Boardman JD. Polygenic influence on educational attainment: new evidence from the National Longitudinal Study of Adolescent to Adult Health. AERA Open. 2015;1:1–13. doi: 10.1177/2332858415599972 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Domingue BW, Conley D, Fletcher J, Boardman JD. Cohort effects in the genetic influence on smoking. Behav Genet. 2016;46:31–42. doi: 10.1007/s10519-015-9731-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Dudbridge F. Power and predictive accuracy of polygenic risk scores. PLoS Genet. 2013;9:e1003348. doi: 10.1371/journal.pgen.1003348 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Juonala M, Magnussen CG, Berenson GS et al. Childhood adiposity, adult adiposity, and cardiovascular risk factors. N Engl J Med. 2011;365:1876–1885. doi: 10.1056/NEJMoa1010112 [DOI] [PubMed] [Google Scholar]
  • 43. Zhu XC, Tan L, Wang HF et al. Rate of early onset Alzheimer’s disease: a systematic review and meta-analysis. Ann Transl Med. 2015;3:38. doi: 10.3978/j.issn.2305-5839.2015.01.19 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. HealthyPeople.gov. Healthy People 2020. Washington, DC: U.S. Department of Health and Human Services. [Google Scholar]
  • 45. Domingue BW, Belsky DW, Harrati A, Conley D, Weir D, Boardman J. Mortality selection in a genetic sample and implications for association studies. bioRxiv. 2016:049635. doi: 10.1093/ije/dyx041 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Tables S1-S11

Articles from The Journals of Gerontology Series A: Biological Sciences and Medical Sciences are provided here courtesy of Oxford University Press

RESOURCES