Childhood Misfortune as a Threat to Successful Aging: Avoiding Disease

Abstract

Purpose:

The purpose of this study was to examine whether childhood misfortune reduces the likelihood of being disease free in adulthood.

Design and Methods:

This article used a sample of 3,000+ American adults, aged 25–74, who were first interviewed in 1995 and reinterviewed in 2005. Logistic regression was used to estimate the odds of avoiding disease at the first wave and remaining disease free a decade later.

Results:

Consistent with a life course view of successful aging, higher levels of childhood misfortune (e.g., abuse, financial strain) are associated with a lower probability of disease avoidance. This pattern was observed across a large set of chronic conditions and in multivariate analyses spanning both waves of the study.

Implications:

Childhood misfortune has approximately equal consequences for adult disease avoidance as does the combined effect of moderate lifetime smoking and obesity. Efforts to alleviate adverse experiences for children may have long-term benefits for successful aging.

Gerontology’s successful aging paradigm has attracted considerable interest over 25 years. Ever since seminal writings by Williams and Wirths (1965) and Rowe and Kahn (1987), scholars have clarified, extended, and reinvigorated the fundamental insight that age is not associated with intrinsic and monotonic losses in physical functioning (Baltes & Carstensen, 1996; Halfon & Hochstein, 2002). Rather, social and behavioral conditions can slow or modify trajectories of decline and prolong healthy life. Current descriptions of successful aging view it as consisting of multiple dimensions—avoidance of disease and disability, high cognitive and physical functioning, and engagement with life—reflected by both objective and subjective benchmarks (Rowe & Kahn, 1997; Strawbridge, Wallhagen, & Cohen, 2002).

In recent years, scholars have incorporated insights from life course epidemiology into the successful aging paradigm, showing that conditions from early adulthood and even childhood contribute to the aging process (Britton, Shipley, Singh-Manoux, & Marmot, 2008; Pruchno, Wilson-Genderson, Rose, & Cartwright, 2010). The present article follows this line of analysis but considers a broader array of early-life conditions than were available in most prior studies. In particular, we build on emerging evidence suggesting that childhood misfortune accumulated across various life domains (family dissolution, abuse and maltreatment, and financial strain) leaves an enduring stamp on well-being across the life course (Schafer, Ferraro, & Mustillo, 2011).

This study focuses on a key aspect of successful aging: avoiding disease. In some ways, remaining disease free is foundational because it is correlated with and enables second-order components such as physical and cognitive function and active engagement with life. Indeed, Rowe and Kahn (1997) articulated that “the relationship among them [components of successful aging] is to some extent hierarchical” (pg. 433).

Another distinctive element of this study is our attention to adults across a wide age range. The data from which our empirical analyses are drawn is the Midlife Development in the United States (MIDUS) Study, a project intended to chart well-being among America from the ages of 25–74, the long stretch of time thought of as “middle age” (Brim, Ryff, & Kessler, 2004). It is useful to examine successful aging among older adults only, but one must recognize that early disease onset may lead to early mortality; thus, studying older people only may mask some of the processes leading to seriously compromised health trajectories. Therefore, we see studies of a wide age range as an opportune way to examine whether childhood misfortune leads to the premature presence of chronic disease and the onset of comorbidity. Moreover, by tracking these subjects over a decade, one can observe some elements of successful aging up to 84 years of age.

Early Adversity, Getting Disease, and Avoiding Disease

Linking childhood conditions, especially adverse conditions, to increased risk of diseases in middle and older age is fairly common practice among epidemiologists (Lynch & Smith, 2005), demographers (Hayward & Gorman, 2004), and other life course scholars (Ferraro & Shippee, 2009). Preston, Hill, and Drevenstedt (1998) cogently summarize the viable underlying mechanisms linking childhood or adolescence with chronic disease: Early misfortune can heighten risk of future health problems directly (through stress or other physiological mechanisms) or indirectly (through the continuity of adverse environments or initiation of poor health behaviors).

Prior studies provide support for both direct and indirect effects of childhood misfortune on health. A growing literature documents that disadvantages such as low socioeconomic status (SES), abuse and maltreatment, family disruption, and health problems during childhood induce later risk of mortality (Hayward & Gorman, 2004), chronic disease (O’Rand & Hamil-Luker, 2005), poor self-rated health (Irving & Ferraro, 2006), and biological inflammation (Slopen et al., 2010), even after adjusting for adulthood conditions such as SES and marital status. Several studies, on the other hand, find that the effects emerging during adulthood are indirect, whereby the relationship between early adversity and adult health is reduced after controlling for the intermediate variable. For instance, Blackwell, Hayward, and Crimmins (2001) reported that the effect of childhood disease on cancer prevalence was partially mediated by adult wealth. Langenberg, Kuh, Wadsworth, Brunner, and Hardy (2006) discovered partial mediation when they found that the effect of childhood social class on metabolic syndrome in women was reduced after controlling for adult social class.

Against this backdrop, applying the successful aging model highlights an alternative approach. Determining which people are at risk of particular diseases is not the same as identifying who is able to avoid any diseases. Being disease free in middle age is important for successful aging, but being disease free in later life is exceptional. Older adults with no chronic disease are, in many respects, an elite group—and, importantly, one that appears to be shrinking over time (McLaughlin, Connell, Heeringa, Li, & Roberts, 2010) Thus, we ask whether childhood misfortune reduces the likelihood of being disease free in adulthood: Does childhood misfortune compromise elements of successful aging?

Many studies on childhood conditions and adult health focus on specific life-threatening diseases, typically identifying how an early event or condition raises the relative risk of condition X, net of more proximal circumstances and demographic factors. The preoccupation with major diseases such as cardiovascular disease or cancer is understandable from a public health standpoint, insofar as the most serious conditions deserve the most sustained research attention. Yet despite their status as the two leading causes of death in the United States, prevalence statistics indicate that 92% and 96% of American adults do not have coronary heart disease or cancer, respectively (American Cancer Society, 2010; American Heart Association, 2009). Avoiders of heart disease may nevertheless be plagued by bone or joint discomfort, dental problems, severe allergic reactions, or a host of other diffuse conditions, which do not pose immediate risk of mortality but compromise quality of life. Dodging the major killers, then, is in many ways a necessary but insufficient condition for successful aging.

When the interest is on how people avoid disease altogether—rather than contract-specific conditions—attention shifts away from pinpointing specific intervening mechanisms that cause disease. Across an aggregate of diseases, it may be less important to uncover which particular pathway variable is important for specific condition A, for specific condition B, and so on, than to invert the question and consider what common factor underlies avoidance of all conditions. The issue becomes more acute when conditions A, B, … K are rare or have discrepant underlying etiology.

Another common tactic in the study of early disadvantage and adult health is to study very general outcomes, such as self-reported health (Irving & Ferraro, 2006). Although many of these studies are illuminating, there is an overlap between physical and mental health inherent in global self-health evaluations. In order to isolate and target disease avoidance, we focus on the prevalence and incidence of a wide range of conditions across different bodily systems.

Research Questions

To guide the analysis to follow, we articulate four specific research questions:

• 1) Does childhood misfortune reduce the likelihood of being disease free, regardless of the condition’s seriousness, etiology, or target organ?

• 2) After adjusting for demographic characteristics, adult socioeconomic status, and other correlates of successful aging, does childhood misfortune decrease the probability of being disease-free in adulthood?

• 3) Using longitudinal data, does childhood misfortune decrease the probability of remaining disease-free at a 10-year follow-up?

• 4) Are the effects of childhood misfortune on disease avoidance more or less consequential according to age?

We hypothesize that childhood misfortune reduces the likelihood of being disease free, both in cross-sectional and longitudinal analyses and after adjusting for relevant covariates. We also anticipate that the effect of childhood misfortune on being disease free will span all ages but will be stronger at younger ages than in later life. This expectation is congruent with the concept of “usual” aging—physiologic loss and increased susceptibility to a variety of diseases is a basic corollary of older age (Rowe & Kahn, 1987). Accordingly, childhood misfortune may pose a downward extension of age-related morbidity. Those with a more auspicious early life, however, are likely more successful in forestalling the encroachment of chronic disease, if not avoiding it altogether.

Design and Methods

Participants

This study uses data from the MIDUS Study, a national probability sample of 3,032 noninstitutionalized Americans aged 25–74 in 1995. Respondents were selected from a random-digit-dialing sample (response rate = 70%). Participants were then given a two-part follow-up questionnaire, yielding an 86.6% response rate. Thus, the overall response rate was 61% (.70 × .87 = .61). Respondents from Wave 1 (W1) were contacted about ten years later to secure their participation for Wave 2 (W2) in 2005. A total of 245 W1 respondents died before the W2 survey. From the remaining W1 respondents, 1,748 individuals completed both the telephone and self-administered follow-up interview (resulting in a W2 response rate of 63%; American Association for Public Opinion Research, 2011). Of the original 3,032, 123 cases were removed for item-missing data at the initial wave, resulting in a final sample of 2,909 for multivariate W1 analyses. For W2, 57 of the 1,748 tracked cases were missing morbidity data and were thus excluded from relevant analyses.

Measures

The MIDUS study contained a wide range of questions related to disease morbidity, and we capitalize on these measures to construct our indicators of disease avoidance at W1 and W2. In the course of the telephone interview, MIDUS respondents were asked “Have you ever had heart trouble suspected or confirmed by a doctor?” and later, “Have you ever had cancer?”. The bulk of the morbidity information was gleaned from the self-administered questionnaires in which participants were asked to indicate whether they had “experienced or been treated for” 29 separate conditions during the past year. The 31 diseases assessed in MIDUS were then categorized according to their International Classification of Diseases 9 (ICD-9) diagnostic codes. Because we focus on physical morbidity, conditions classified as mental disorders were removed ([a] “anxiety, depression, or some other emotional disorder”; [b] “alcohol or drug problem”; and [c] “chronic sleeping problems”).

Indicators of disease avoidance were created from the morbidity information available at W1 and W2. For both waves of the study, a binary variable differentiates those who were disease free (coded 1) from those who had at least one condition (coded 0).

The primary independent variable of this study, childhood misfortune, has had multiple operationalizations in the literature. We follow the approach of Felitti and colleagues (Edwards, Holden, Felitti, & Anda, 2003; Felitti et al., 1998) who have advanced the utility of an additive adversity measure. In essence, this tactic sums across different types of misfortune in order to capture the “cumulative burden of multiple traumas” (Turner & Lloyd, 1995, p. 268). For the current study, we utilized a host of retrospective childhood questions available in the MIDUS data. These items include (a) receipt of welfare; (b) less than a high school education for father (or mother, in households where father was not present); (c) report of being “worse off” than other families; (d) lack of male in the household; (e) parental divorce; (f) death of a parent; physical abuse at the hands of a (g) mother, (h) father, (i) siblings, or (j) other person; (k–n) emotional abuse by any of the same parties; (o) fair or poor physical health at age 16; and (p) fair or poor mental health at age 16.

Following Felitti and colleagues (1998), each indicator was coded in dichotomous form, reflecting a major insult during childhood (e.g., responses to the eight abuse items ranged from never [coded 1] to often [coded 4], and abuse was coded as present if reported as sometimes or often). In order to retain as many respondents as possible, we generated an average score for all participants with at least half of the questions answered. Childhood misfortune refers to the count of events, experiences, and disadvantaged positions during childhood.

In addition to childhood misfortune, multivariate analyses include a number of statistical controls for demographic characteristics and conceptually important covariates. The goal is to examine whether childhood misfortune threatens disease avoidance above and beyond other elements or correlates of successful aging commonly found in the literature and available in the MIDUS study (c.f. Rowe & Kahn, 1997). We include four indicators of variables known to heighten risk of multiple diseases: obesity (1 = body mass index [BMI] ≥ 30, 0 = otherwise), smoking (estimated number of lifetime cigarettes/100,000), heavy drinking (1 = consuming more than 3 alcoholic drinks a day, 0 = otherwise), and sedentary lifestyle (1 = engaging in vigorous or moderate activity less than once a month, 0 = otherwise).

Social support was operationalized as the average response to four questions concerning the supportiveness of their family relations and four questions about their friend relations (e.g., “How much do they understand the way you feel about things?”). Valid responses range from 1 (not at all) to 4 (a lot), and both indexes demonstrate acceptable reliability (α = .82 for family relations and α = .88 for friend relations).

Personal efficacy was assessed with a 12-item index asking respondents to assess statements (1 = strongly disagree to 7 = strongly agree) such as the following: “In general I feel I am in charge of the situations in which I live.” Taken together, the items had high reliability (α = .85).

Two variables were used to tap productivity in life, the first a basic indicator of work status (1 = currently working for pay, 0 = otherwise). The second variable is mean response to the six-item Loyola Generativity Scale, which assesses a broader sense of meaningful life productivity. Respondents were asked to assess the degree to which statements such as “You feel that other people need you” were accurate (1 = not at all to 4 = a lot). Jointly, the six items constitute a reliable index (α = .84).

Education level is a pivotal component of adult SES and is measured by the approximate number of years of schooling attained.

Finally, a number of basic demographic variables are included in the analysis, including age (year of birth subtracted from 1995), gender (1 = female, 0 = male), race (1 = non-White, 0 = White), and whether respondent lives alone (1 = yes, 0 = otherwise).

Results

The initial aim of this article is to examine whether early misfortune reduces the likelihood of being free of disease, and we examine a wide—and, perhaps, seemingly unconnected—set of conditions. Table 1 provides a straightforward comparison of those who were disease free and those who had the condition at W1, profiling their mean scores on childhood misfortune. Although disease prevalence is infrequent for some conditions, we provide two-tailed t tests of means on the childhood misfortune scores. The 28 conditions are organized by their ICD-9 coding (Centers for Disease Control and Prevention, 2011). The MIDUS indicators for disease span 11 ICD-9 categories and attest to the diversity of diseases represented in the analysis. What is most notable from Table 1, then, are the consistently lower levels of childhood misfortune among those who are disease free. In 27 of 28 conditions, disease avoiders have lower childhood misfortune scores than those with the particular disease. In some cases, the difference is quite small and statistically nonsignificant with a t test (e.g., stroke victims have average misfortune score of 2.65 vs. 2.64 among stroke avoiders). For other conditions, however, the disparity is more marked (e.g., misfortune scores of 3.41 vs. 2.63 for varicose veins, p < .05). It is worth bearing in mind that some conditions represented in Table 1 are quite rare (e.g., AIDS, tuberculosis); thus, probability levels are provided for .05 and .10. Regardless of the size of the difference for each of the 28 specific diseases or the condition’s unique prevalence, the scope and pattern of the differences is remarkable when considering the portfolio of diseases.

Table 1.

W1 Prevalence of Various Conditions Grouped by ICD-9 Codes and Mean Childhood Misfortune Score in the Midlife Development in the United States Study

Disease	Mean of childhood misfortune score (number of cases)
	Has disease (number with disease)	Disease free (number without disease)
Infectious and parasitic diseases (ICD9 Code 1)
Tuberculosis	2.71 (7)	2.64 (3,003)
AIDS/HIV	3.50 (10)	2.64 (3,009)
Neoplasms (ICD9 Code 2)
Cancer	2.74 (212)	2.63 (2,816)
Endocrine, nutritional and metabolic diseases, and immunity disorders (ICD9 Code 3)
Thyroid disease	2.79 (134)	2.63 (2,873)
Diabetes	2.92 (165)	2.61 (2,851)*
Diseases of the nervous system and sense organs (ICD9 Code 6)
Multiple sclerosis, epilepsy, or other neurological problem	3.04 (57)	2.64 (2,956)
Migraine headaches	3.14 (319)	2.59 (2,698)**
Diseases of the circulatory system (ICD9 Code 7)
Varicose veins requiring treatment	3.41 (41)	2.63 (2,973)**
Hypertension	2.63 (555)	2.64 (2,457)
Stroke	2.65 (31)	2.64 (2,983)
Piles or hemorrhoids	3.06 (344)	2.59 (2,671)**
Heart problem	2.75 (377)	2.62 (2,647)
Diseases of the respiratory system (ICD9 Code 8)
Hay fever	2.80 (493)	2.61 (2,518)*
Diseases of the digestive system (ICD9 Code 9)
Hernia or rupture	3.03 (100)	2.62 (2,913)*
Recurring stomach trouble, indigestion, or diarrhea	3.11 (606)	2.52 (2,405)**
Constipation	3.06 (195)	2.62 (2,815)**
Gallbladder trouble	3.28 (70)	2.63 (2,948)**
Gum or mouth problems	3.22 (227)	2.60 (2,787)**
Teeth problems	3.20 (288)	2.58 (2,719)**
Diseases of the skin and subcutaneous tissue (ICD9 Code 12)
Persistent skin trouble (e.g., eczema)	2.99 (332)	2.60 (2,678)**
Ulcer	3.25 (130)	2.62 (2,883)**
Diseases of the musculoskeletal system and connective tissue (ICD9 Code 13)
Arthritis, rheumatism, or other bone or joint diseases	2.74 (605)	2.62 (2,400)
Sciatica, lumbago, or recurring backache	2.93 (620)	2.58 (2,385)**
Foot trouble (e.g., bunions, ingrown toenails)	3.15 (355)	2.57 (2,657)**
Lupus or other autoimmune disease	2.69 (39)	2.64 (2,974)
Congenital anomalies (ICD9 Code 14)
Urinary or bladder troubles	3.03 (408)	2.58 (2,595)**
Chronic obstructive pulmonary disease and allied conditions
Asthma, bronchitis, or emphysema	2.98 (395)	2.59 (2,617)**
Other lung problem	3.48 (115)	2.61 (2,887)**

Notes: ICD9 = International Classification of Diseases 9.

^*p < .1 (t test between has disease and disease-free groups).

^**p < .05 (t test between has disease and disease-free groups).

Having shown that for most conditions assessed in the MIDUS study, disease avoiders had lower levels of childhood misfortune, the next step is to assess whether the pattern holds for remaining completely disease free from all conditions. Table 2 presents the results of two binary logistic regressions, which address this question at baseline and at W2. Baseline analyses consist of the entire analytic sample, whereas W2 analyses are estimated among only those persons who were disease free at W1. Taken together, the analyses allow us to identify who is more likely to avoid disease and, among those who have successfully eluded disease at baseline, who is able to sustain this track record of successful aging over the course of an entire decade. The W1 and W2 analyses, therefore, refer to the inverse of disease prevalence and incidence. Both models include all covariates described above.

Table 2.

Odds Ratios (and 95% confidence intervals) From Logistic Regression of W1 and W2 Analyses, Midlife Development in the United States Study

	Disease free, W1	χ² Test for age differences (under 50 vs. 50+)	Disease free, W2	χ² Test for age differences (under 50 vs. 50+)
Childhood misfortune	0.93* (0.88–0.99)	No age differences (p = .25)	0.88* (0.77–1.00)	No age differences (p = .90)
Covariates
Lifetime smoking	0.93* (0.87–1.00)	No age differences (p = .57)	0.82 (0.66–1.01)	No age differences (p = .99)
Obese	0.67* (0.51–0.87)	No age differences (p = .34)	0.46* (0.24–0.85)	No age differences (p = .24)
Heavy drinking	0.83 (0.53–1.30)	No age differences (p = .88)	1.03 (0.43–2.51)	No age differences (p = .83)
Sedentary lifestyle	0.32 (0.09–1.08)	No age differences (p = .25)	a
Social support family	0.97 (0.81–1.17)	No age differences (p = .28)	1.15 (0.72–1.84)	No age differences (p = .80)
Social support friends	1.11 (0.93–1.31)	No age differences (p = .43)	0.83 (0.55–1.26)	No age differences (p = .20)
Personal efficacy	1.31* (1.16–1.47)	No age differences (p = .97)	1.07 (0.80–1.44)	No age differences (p = .36)
Working	1.16 (0.87–1.56)	No age differences (p = .49)	0.89 (0.46–1.71)	No age differences (p = .98)
Sense of productivity	0.96 (0.79–1.15)	No age differences (p = .54)	1.00 (0.64–1.57)	No age differences (p = .29)
Education	0.98 (0.94–1.02)	No age differences (p = .55)	1.03 (0.94–1.14)	No age differences (p = .74)
Age	0.97* (0.96–0.98)		0.97* (0.95–0.99)
Non-White	1.16 (0.84–1.61)	No age differences (p = .30)	0.66 (0.28–1.53)	No age differences (p = .42)
Female	0.70* (0.56–0.87)	Stronger effect among older group (p = .002)	0.83 (0.50–1.39)	No age differences (p = .94)
Live alone	1.12 (0.86–1.45)	No age differences (p = .63)	0.81 (0.46–1.44)	No age differences (p = .89)
−2LL	1509.68		248.52
N	2,908		390b

Notes: Childhood misfortune = count of 16 potential misfortunes related to abuse and maltreatment, financial strain, family instability, health problems; lifetime smoking = estimated number of lifetime cigarettes/100,000; obese = body mass index > 30; heavy drinking = reporting more than 3 drinks per day; sedentary lifestyle = engaging in vigorous or moderate activity less than once a month; social support family = four-item index of questions concerning supportiveness of family relations; social support friends =four-item index of questions concerning supportiveness of friend relations; personal efficacy = 12-item index concerning control over life circumstances; working = currently working for pay; sense of productivity = six-item index concerning engagement in productive life activity; education = years of completed education; age = 1995—year of birth; non-White = racial classification (White is reference group); female = gender of respondent (male is reference group); live alone = household composition (lives with others is reference group).

^aEffect of sedentary lifestyle could not be estimated for W2 analyses because an insufficient number of respondents were sedentary and disease free at W1.

^bOf respondents disease free at W1 and reinterviewed at W2.

^*p < .05

Odds ratios displayed in Table 2 indicate that for each unit increase in the childhood misfortune score, the probability of being disease free at W1 decreases by 6.7% (first column). This decrease is net of demographic factors as well as other important dimensions of successful aging, including health-related behaviors (e.g., smoking), engagement in activity (e.g., currently working), social support, and self-efficacy. Whereas the data reveal that childhood misfortune reduces the likelihood of being disease free, we also examined whether this relationship varied by age; the sample was divided into two groups (<50 and ≥50), and ancillary analyses considered additional age cutoffs but yielded conclusions consistent with those presented below.

Potential age differences in the model were assessed with a Wald χ² test using Hoetker’s (2007)—complogit—routine in Stata, which tests for subgroup variability while accounting for residual variance (unobserved heterogeneity) in nonlinear equations (Allison, 1999). Comparing the effect sizes of each coefficient (log of the odds ratios presented in Table 2) across age subgroups, the only observed difference is for the female covariate. Specifically, the probability of disease avoidance is substantially lower among older women, but gender differences are null among people younger than 50 years. As far as childhood misfortune is concerned, the lack of difference across age groups suggests that childhood misfortune is equally consequential on being disease free across the two age groups.

Moving to the analysis for W2, the results further confirm the challenge that childhood misfortune poses to disease avoidance. Among the relatively healthy subgroup, who managed to dodge disease at W1, the odds of remaining disease free became 12% less likely for each unit increase in childhood misfortune.

The only other variables that had a consistent statistically significant association with disease avoidance at baseline and at W2 were obesity and age. Obese MIDUS respondents were 32% less likely than non-obese adults to avoid disease at W1. Were they able to sidestep the 28 conditions at W1, obese individuals still faced 55% lower odds of remaining disease free by W2 relative to their non-obese counterparts. Each additional year of age was associated with a 3% decrease in the odds of avoiding disease, both at baseline and at the follow-up. Testing for interactions by age, however, revealed that the effect of childhood misfortune on being disease free at W2 was nonsignificant.

Discussion

According to many gerontologists, avoiding disease is a crucial element of successful aging. Even if not essential to successful aging, being disease free after age 60, 70, or 80 is remarkable. Although by middle age it becomes commonplace to confront some form of chronic condition, a subset of exceptionally healthy individuals is able to navigate through adulthood unscathed by life-threatening diseases and bothersome conditions. In this article, we examined whether childhood misfortune lowers the likelihood of attaining the elite benchmark of disease avoidance. Genetics, good health behaviors, and supportive social relations are understood to play an important role in remaining disease free (Rowe & Kahn, 1997), but research on successful aging has only begun to explicitly consider early-life conditions. To that end, we incorporated a set of retrospective questions about abuse, family structure, financial hardship, and health challenges to test whether the long arm of childhood (Hayward & Gorman, 2004) reaches to the prospects of disease avoidance and successful aging.

The results of this study corroborate the claim that childhood is a sensitive period of life course development (Lynch & Smith, 2005), as adverse events and conditions compromise disease avoidance in middle age and beyond. The long arm of childhood was reflected in several overlapping ways. First, greater exposure to childhood misfortune was associated with a lower probability of sidestepping a broad range of health problems. To put this finding in quantitative context, a non-smoking, non-obese 50-year-old male who faced six distinct indicators of misfortune (the 90th percentile of misfortune in our data) would be expected to attain disease avoidance with 21.4% probability. In contrast, the predicted probability of being disease free for that identical person under the condition of no childhood adversities would be 29.1%. Had such a no-misfortune adult smoked moderately (25th percentile among ever-smokers or 60th percentile of the overall sample) and been obese (BMI ≥ 30), his probability of disease avoidance would be 20.7%. Though women, on average, are less likely to remain free of chronic diseases, they face consequences proportionate to men. To illustrate, Figure 1 presents these predicted probabilities. The similarity in effect between childhood misfortune and these established health factors testifies to a needed life course emphasis in successful aging scholarship. Clearly, early-life course experiences set the tone for future life trajectories (Schafer et al., 2011) and should not be glossed over in search of proximal adult risk factors only. The effect of childhood misfortune on being disease free during adulthood is not trivial; the effect size is on par with widely known risk factors.

Figure 1.

Comparison of childhood misfortune with other important risk factors for the predicted probability of disease avoidance. Note: Probabilities are calculated from the multivariate logistic regression model presented in Table 1. Besides childhood misfortune, smoking, obesity, age, and gender, all control variables are held at their means. Moderate smoking is defined as the 25th percentile lifetime smoking value for people who have ever smoked in the sample (equivalent to nearly 66,000 lifetime cigarettes; 65% of MIDUS respondents within that approximate value were former smokers in the 1995 interview).

Second, the impact of childhood misfortune lingered onward through a decade of observation. Even among healthy adults who exemplified the disease-free side of successful aging, childhood misfortune remained a threat to disease avoidance at the follow-up wave of the MIDUS Study. We expected stronger effects during the early part of the adult life course but found that the effect of childhood misfortune on being disease free extended well into later life. To use the metaphor of an athletic tournament, individuals bearing the stamp of early accumulated misfortune may not be eliminated in the initial round of competition; but their likelihood of advancing to the elite stages of the game becomes increasingly slim. To be sure, few adults manage to completely sidestep chronic morbidity through the course of middle age; in the MIDUS Study, only 176 of the initial 3,000+ person sample were disease free at W1 and remained in this privileged class by W2. But childhood misfortune compounds the unlikeliness of the feat, even while other correlates of successful aging waned in their influence.

This article adds to the emerging body of literature linking early-life conditions to the successful aging process (Britton et al., 2008; Pruchno et al., 2010). Whereas past epidemiological research on health and early adversity focuses on specific conditions (e.g., heart disease, O’Rand & Hamil-Luker, 2005 or cancer, Fuller-Thompson & Brennenstuhl, 2009), the current study draws from the premise that successful aging is not limited to any isolated ailment and points to the utility of examining diseases across a wide range of bodily systems. In entering this dialogue, several contributions and limitations of the present study become evident.

Whereas prior research has tended to conceptualize early-life conditions as some combination of educational attainment, father’s social class, adult incarceration, or even basic demographic traits such as gender or race (Britton et al., 2008; Pruchno et al., 2010), the current study incorporated a wide set of events and conditions from childhood. It is surely important to trace the roots of successful aging to transitions in early adulthood (e.g., educational attainment), but we argue that childhood represents a more basic life stage that merits attention in its own right (Ferraro & Shippee, 2009; McLeod & Almazan, 2003). Of course, prospective studies of childhood to late-life are prohibitively costly and time consuming, which often leaves retrospective study designs the next-best alternative. Relying on participants’ recollections about childhood information necessarily raises concerns about forgotten details and respondent bias. We acknowledge this as a limitation of the study design. Past research using a similar approach suggests that underestimation tends to be more common than overestimation (i.e., people report fewer adversities than occurred), and this often leads to downwardly biased results (Dube, Felitti, Dong, Giles, & Anda, 2003).

Another potential limitation of this article is that it examines only one aspect of successful aging: avoiding disease. Indeed, much of successful aging’s attraction lies in its multidimensionality (Rowe & Kahn, 1997). At the same time, the narrow focus of our inquiry offers several distinct contributions. For instance, careful attention to disease actually reveals much about the scope and breadth of misfortune’s influence. Complete disease avoidance is a remarkable endeavor, in part because there are dozens of unrelated conditions a person could theoretically contract. Among infectious and parasitic diseases (ICD-9 Code 1); neoplasms (ICD-9 Code 2); endocrine, nutritional, and metabolic diseases, and immunity disorders (ICD-9 Code 3); and on down the line of potential health problems, disease avoiders had consistently lower levels of early misfortune. Perhaps just as important, many conditions examined in this study—such as diabetes, persistent skin trouble, varicose veins, and diseases of the digestive system—are not the outcomes of a solitary proximal risk factor in adulthood that childhood misfortune is likely to induce. Rather, the association between early-life misfortune and adult disease reveals the working of a metamechanism—a fundamental risk factor producing an extensive array of subsequent mechanisms, each disproportionately threatening health (Freese & Lutfey, 2011). Proactive protection from childhood misfortune can therefore have a broad influence on people’s well-being in later life, in part by minimizing the probability of many and unrelated chronic diseases. Disease avoiders, regardless of the conditions considered, are not likely to be those with a poor start in life.

Finally, a major finding of this article, that early-life events are consequential for disease avoidance, leads to logical next steps of investigation. Future research can target other important elements of adult well-being, such as physical and cognitive functioning, active engagement in purposeful tasks, and social relations with friends and family. The current analysis examined one component of successful aging, albeit a pivotal one to most models of successful aging.

Funding

Support for this research was provided by a grant from the National Institute on Aging (R01 AG033541).