Developmental Psychology 2003, Vol. 39, No. 5, 924 –933
Copyright 2003 by the American Psychological Association, Inc. 0012-1649/03/$12.00 DOI: 10.1037/0012-16126.96.36.1994
A Multimethodological Analysis of Cumulative Risk and Allostatic Load Among Rural Children
Gary W. Evans
This study merged two theoretical constructs: cumulative risk and allostatic load. Physical (crowding, noise, housing quality) and psychosocial (child separation, turmoil, violence) aspects of the home environment and personal characteristics (poverty, single parenthood, maternal high school dropout status) were modeled in a cumulative risk heuristic. Elevated cumulative risk was associated with heightened cardiovascular and neuroendocrine parameters, increased deposition of body fat, and a higher summary index of total allostatic load. Previous findings that children who face more cumulative risk have greater psychological distress were replicated among a sample of rural children and shown to generalize to lower perceptions of self-worth. Prior cumulative risk research was further extended through demonstration of self-regulatory behavior problems and elevated learned helplessness.
Theoretical advances in the understanding of the ecology of human development (Bronfenbrenner & Evans, 2000; Bronfenbrenner & Morris, 1998; Wachs, 2000) have not been matched by analytical methods for capturing the dynamic, systemic interplay of the organismic, proximal, and distal processes that shape development. Statistical interaction formulations are inadequate to capture the ecology of human development. Interactions, particularly higher order terms, have low statistical power. Moreover, researchers’ ability to comprehend high-level (i.e., greater than three-way) interactions is limited. Twenty years ago, Michael Rutter (1983) suggested an alternative approach to the analysis of complex systems operative in human development. In Rutter’s approach, organismic characteristics as well as proximal and distal qualities of the social and physical environment are modeled collectively in what have come to be called cumulative risk models. For each person or environment construct, a dichotomous classification of risk exposure is determined, typically by a statistical cutoff (e.g., greater than one standard deviation above the mean, upper quartile) or on the basis of a conceptual categorization (e.g., being below the poverty line, single parenthood). Cumulative risk is then calculated by a simple summation of the multiple risk categories (Rutter, 1983, 1993).
Partial support for this project came from the John D. and Catherine T. MacArthur Foundation Network on Socioeconomic Status and Health, the W. T. Grant Foundation, the Cornell University Agricultural Experiment Station (Projects NYC 327404 and NYC 327407), and the National Institute of Child Health and Human Development (Grant 1 F33 HD08473– 01). I am very grateful to the children and families who are participating in this research program. This work has benefited enormously from the sage counsel of Urie Bronfenbrenner and the able assistance of Jana Cooperman, Kimberly English, Missy Globerman, and Amy Schreier in data collection. I thank Lyscha Marcynyszyn and Nancy Wells for critical feedback on an earlier version of this article. Correspondence concerning this article should be addressed to Gary W. Evans, Department of Design and Environmental Analysis and Department of Human Development, Cornell University, Ithaca, New York 14853– 4401. E-mail: firstname.lastname@example.org 924
The chief advantage of Rutter’s (1983, 1993) cumulative risk metric is its ability to simultaneously model a large number of risk factors without the major statistical and interpretation liabilities of multiplicative interactions. Cumulative risk models also reflect the typical, natural covariation of many childhood risk factors. Sociocultural forces such as poverty and racism tend to allocate risk disproportionately in societies to subsets of the population such as the poor and ethnic minorities. Thus, a concentration of physical and social risks is often focused on the most vulnerable population strata in many cultures (Schell, 1997). For example, physical risk factors such as poor housing quality, noise, and pollution are highly intercorrelated (Evans & Kantrowitz, 2002), as are psychosocial risk parameters such as poverty, family turmoil, and exposure to violence (Lepore & Evans, 1996; Rutter, 1993). Another advantage of cumulative risk modeling, initially unforeseen in developmental theory, is its theoretical convergence with recent advances in neurobiology and stress, termed allostasis. Allostasis, in contrast to homeostasis, posits a dynamic, highly interactive set of multiple physiological systems of equilibrium maintenance with which the body continuously adjusts its normal operating range (set points) in response to multiple physical and social demands (McEwen, 1998; McEwen & Stellar, 1993; Seeman & McEwen, 1996; Sterling & Eyer, 1988). These dynamic adjustments reflect downward regulation in order to maintain the organism’s internal stability but at levels more congruent with environmental demands. The active, ongoing maintenance of internal equilibrium increases allostatic load, which reflects chronic wear and tear on the body caused by the mobilization of resources to meet changing environmental demands (McEwen, 1998, 2000; McEwen & Seeman, 1999). Overexposure to a combination of neuronal, endocrine, and immune mobilizations alters the ability of the body to respond efficiently to demands (Karlamangala, Singer, McEwen, Rowe, & Seeman, 2002; McEwen, 1998; Seeman, McEwen, Rowe, & Singer, 2001; Seeman, Singer, Rowe, Horwitz, & McEwen, 1997). If exposure to environmental demands is high, the body will mobilize resources to meet these demands but at a higher level of activity. During periods of relative calm, individuals who have a higher allostatic load burden will become less
CUMULATIVE RISK AND ALLOSTATIC LOAD
efficient in turning off the multiple physiological resources marshaled to deal with the prior demands. Allostatic load in the cardiovascular system, for example, is marked by hemodynamics reflecting elevated basal levels of blood pressure, blood pressure hyperreactivity to acute demands, and inhibited blood pressure recovery (i.e., delayed recovery to baseline levels). Hypothalamicpituitary-adrenocortical axis (HPA)-mediated changes in hippocampal architecture, neuroendocrine-induced alterations in lipid metabolism, hemodynamics, and translocation of immune cells are among the hypothesized underlying mechanisms causing such shifts in bodily efficiency for coping with stressors (McEwen, 2000). Alterations in neuroendocrine activity both in the sympathetic nervous system and the pituitary adrenal axis, elevated cardiovascular activity, depressed immune function, heightened fat deposition along with lower HDL/LDL ratios are physiological changes in allostatic load reflecting chronic wear and tear on the body (Karlamangala et al., 2002; McEwen, 1998, 2000; McEwen & Seeman, 1999; Seeman et al., 1997, 2001). Allostatic load is not, however, simply a consequence of environmental demands. Rather, it is a complex, dynamic system of physiological changes in multiple systems created by responses to environmental demands that are modulated by prior experience with stressors, genetic predisposition, and lifestyle choices (e.g., diet, physical activity) (McEwen, 1998; McEwen & Seeman, 1999; Seeman et al., 2001; Seeman, Singer, Ryff, Love, & LevyStorms, 2002). Four important implications of allostasis theory that are of particular relevance to developmental theory are as follows: 1. Allostatic load is cumulative over time, with chronic demands having different and more harmful impacts than short-term stressors. The accumulation of multiple small changes in physiological functioning, rather than specific changes in individual biomarkers, portends eventual physical and psychological morbidity. Seemingly small, modest alterations in functioning in one physiological system when considered jointly with other modest physiological changes may in fact markedly elevate risk, which is otherwise obscured or underestimated when the changes are viewed in isolation. Ideally, allostatic load is marked by the combination of typically modest physiological changes across multiple biological systems in response to demands. Allostatic load is a joint function of physical and social demands throughout life (i.e., current and chronic stressor experience), genetic predispositions, and lifestyle choices such as diet and physical activity. Not only physical morbidity is influenced by elevated allostatic load over the life course. Socioemotional and cognitive processes may be impacted as well. Particularly sensitive processes are hypothesized to include depression, anxiety, and self-regulatory behavior as well as selective attention and spatial and episodic memory (McEwen, 2000; McEwen & Seeman, 1999; Seeman et al., 2001).
The developmental implications of allostasis theory are further examined in the Discussion section. An important drawback of the cumulative risk perspective is its implicit assumption that personal, psychosocial, or environmental risk factors are interchangeable and can simply be added together to represent the total load placed on the organism. As a partial response to this criticism, several investigators have shown that although individual risk factors do in fact vary in their respective impacts, each of these unique effects pales in comparison to the explanatory power of the cumulative risk metric in explaining socioemotional development (Ackerman, Izard, Schoff, Youngstrom, & Kogos, 1999; Liaw & Brooks-Gunn, 1994; Rutter, 1983, 1993; Sameroff, 1998). Parallel results have been shown in the cardiovascular and neuroendocrine systems (McEwen, 1998, 2000; McEwen & Seeman, 1999; Seeman & McEwen, 1996) and in diseases directly related to these processes (Karlamangala et al., 2002; Seeman et al., 1997, 2001). Singular risk factors do not predict socioemotional development, physical morbidity, or underlying physiological processes as well as cumulative risk models do. This article brings together these two lines of inquiry on cumulative risk and allostatic load by examining psychophysiological processes indicative of allostatic load in relation to cumulative risk. I also examined two important socioemotional behaviors potentially linked to allostatic load: self-regulatory behavior and learned helplessness. One of the consequences of chronically elevated neuroendocrine activity, particularly in the HPA, may be impaired self-regulatory ability (McEwen, 1998, 2000; McEwen & Stellar, 1993; Metcalf & Mischel, 1999; Repetti, Taylor, & Seeman, 2002). Difficulties in attention allocation as well as hyperarousal are believed to underlie this connection. In addition, experiences of chronic, multiple demands that are largely intractable heighten susceptibility to learned helplessness (S. Cohen, 1980; S. Cohen, Evans, Stokols, & Krantz, 1986; Glass & Singer, 1972; Peterson, Maier, & Seligman, 1993; Seligman, 1975). Thus, I extended prior developmental research on cumulative risk by incorporating allostatic load measures, behavioral markers of selfregulatory behavior and learned helplessness, along with two questionnaire measures (i.e., self- and maternal ratings) of socioemotional health. All of the previous studies of cumulative risk and child development except one have been restricted to paper-andpencil outcome measures (Ackerman et al., 1999; Barocas, Seifer, & Sameroff, 1985; Dunst & Trivette, 1994; Fergusson, Horwood, & Lynskey, 1994; Furstenburg, Cook, Eccles, Elder, & Sameroff, 1999; Lengua, 2002; Liaw & Brooks-Gunn, 1994; Rutter et al., 1974; Shaw & Emery, 1988; Werner & Smith, 1982; Williams, Anderson, McGee, & Silva, 1990). The only study with multiple methodological outcome measures of cumulative risk that I uncovered was a study showing that multiple risk exposure among preschoolers interfered with self-regulatory behavior (Barocas et al., 1991). A second contribution the present article makes is its focus on a population that has been ignored in the cumulative risk literature— rural children. Despite the fact that there are more high-risk families living in the United States in rural areas than in metropolitan areas (Hernandez, 1993; Sherman, 1992), developmental investigators have focused on inner city populations. To my knowledge, this is the only study of cumulative risk among a rural population. The study of cumulative risk, allostatic load, and socioemotional development is also important because it may shed light on trends
EVANS violence subscales, we used one standard deviation above the mean for the risk categorization. Therefore the total cumulative risk score for each child could vary from 0 to 9. Outcome variables included maternal ratings (0 does not apply; 1 applies somewhat; 2 certainly applies) of psychological symptoms in the target child (anxiety, nervousness, depression, behavioral conduct) on the .83; Boyle & Jones, 1985; Rutter Child Behavior Questionnaire ( Rutter, Tizzard, & Whitmore, 1970). A sample item is “My child often worries, is worried about many things.” In addition, each child evaluated his or her own competency on the Global Self-Worth subscale of the Harter .67; Harter, 1982). A sample item is Perceived Competency Scale ( “Some kids like the kind of person they are” versus “Other kids often wish they were someone else.” After choosing between the two forced alternatives, the child was asked, “Would you say this is sort of true for you or really true for you?” Two behavioral protocols were included to examine self-regulatory behavior and learned helplessness, respectively. The child’s ability to delay gratification was the index of self-regulatory behavior. In this paradigm, a preferred object (a large plate of candy vs. a medium-sized plate of candy) is placed uncovered directly in front of the child. The child is informed that if she or he waits while the experimenter does something in the other room and does not touch the candy or get out of her or his chair, then she or he can have the large plate of candy as soon as the experimenter returns. However, if the child is unable to wait, then she or he can ring a bell placed next to the candy, the experimenter will return immediately, and the child can have the smaller plate of candy. After ensuring that the child understood the instructions, the experimenter left the room until the child either rang the bell or 30 min had elapsed. The session was also videotaped to monitor rule adherence. Mischel and colleagues have shown that this self-regulatory task predicts several concurrent and prospective socioemotional and cognitive outcomes such as behavioral adjustment and school grades (Mischel, Shoda, & Rodriguez, 1989). The number of seconds the child successfully delayed as well as whether she or he delayed for the entire period were recorded. Learned helplessness was evaluated with a standard behavioral protocol developed by Glass and Singer (1972) and adapted for children (Bullinger, Hygge, Evans, Meis, & van Mackensen, 1999; Evans, Hygge, & Bullinger, 1995). The child is given a paper-and-pencil puzzle in which the task is to visit each object (e.g., animal) on the paper by tracing over the interconnecting lines without lifting his or her pencil or doubling back on any line. The child is informed that he or she can keep working on the puzzle by starting over on another copy until he or she solves it or feels unable to. At that point, the child can move on to another puzzle. Once the child has moved to the second puzzle, he or she cannot go back to the first one. An initial practice puzzle is used to make sure the child understands the task. Unbeknownst to the child, the first test puzzle is unsolvable, and the measure of interest is time of persistence.1 All children solve the second puzzle and are assured that they did a good job and that most kids find the first puzzle very difficult. This behavioral protocol is sensitive to individual differences in control-related beliefs, experimental manipulations of perceived control, and chronic exposure to uncontrollable stressors (S. Cohen, 1980; S. Cohen et al., 1986; Evans, 2001; Glass & Singer, 1972). Indicators of allostatic load included resting blood pressure (diastolic and systolic), overnight urinary neuroendocrine measures (cortisol, epinephrine, and norepinephrine), an index of fat deposition (body mass index [kg/m2]), and a total allostatic load index (0 – 6) reflecting the number of these six singular physiological indicators on which each child scored in the top quartile of risk. Prior studies of allostatic load in adults have used similar metrics, combining multiple physiological indicators of risk into one total allostatic load index (Kubzansky, Kawachi, & Sparrow, 1999;
indicative of growing developmental disarray among American children and youth. Many American children are confronted by burgeoning levels of chaos endemic to contemporary family life (Bronfenbrenner, McClelland, Wethington, Moen, & Ceci, 1996; Garbarino, 1995).
Three hundred thirty-nine children (mean age 9.2 years; 49% female; 94% White) were recruited from public schools, New York State CoOperative Extension programs, Head Start programs, and other state and federal programs targeting low-income families in five upstate New York rural counties. The average income-to-needs ratio of the sample was 1.66. An income-to-needs ratio equal to 1 is the federal per capita poverty line. Because this study was part of a larger research program on rural poverty, approximately half of the families were below the poverty line. Families were paid for their participation.
All data were collected with a standardized protocol in participants’ residences. Two interviewers worked independently with the mother and the target child separately. Only 1 child per household was included. Each risk factor was defined dichotomously (0/1) on the basis of statistical or theoretical criteria. Cumulative risk was defined as the simple, unweighted sum of nine risk factors. Individual risk factors were not weighted for three reasons. One, the foundation of cumulative risk theory is that the confluence of risk factors rather than any singular risk, regardless of its particular content, is what leads to dysfunction, because it overwhelms the adaptive capacities of the organism. No one risk factor is seen as essential or more important than another. As noted in the introduction, numerous studies of both physical morbidity and psychological dysfunction support this basic conceptual tenet of cumulative risk theory. Two, weighted predictive models tend to be less reliable than unweighted models, particularly when sample sizes are relatively modest, and they rarely outperform unweighted models over repeated applications (Dawes & Corrigan, 1981). Three, because one of the major objectives in this study was to extend prior developmental research on cumulative risk, it was important that I use the same cumulative risk formulation used in prior studies. Sociodemographic risk factors included poverty (income-to-needs ratio 1), single-parent status, and maternal high school dropout status. Environmental risks included residential crowding (number of people/ room), noise (indoor decibel levels measured over two different, 2-hr periods), and housing problems. Housing problems were assessed by trained raters using a standardized scale that included measures of structural quality, clutter and cleanliness, hazards, indoor climate, and children’s resources (e.g., books, age appropriate toys or games; Evans, Wells, Chan, & Saltzman, 2000). For each of these environmental characteristics, risk was defined as a measurement greater than one standard deviation above the mean for the entire distribution of 339 households in the sample. Psychosocial risk factors included family turmoil, child separation from the family, and exposure to violence. Each of these psychosocial risk factors was assessed by subscales of the Life Events and Circumstances Checklist (LEC; Work, Cowen, Parker, & Wyman, 1990; Wyman, Cowen, Work, & Parker, 1991). Mothers were asked to indicate on multiple dichotomous items (yes/no) whether certain events had occurred during the child’s lifetime. The LEC gauges exposure to multiple chronic stressor domains and has undergone extensive psychometric development. Sample items include “Our child has been involved in serious family arguments,” “A close family member has been away from home a lot,” and “Our neighborhood has been unsafe.” For the family turmoil, child separation, and
Number of attempts on the first puzzle was also measured, and the results were analogous to those for persistence time.
CUMULATIVE RISK AND ALLOSTATIC LOAD Seeman et al., 1997, 2001, 2002; Singer & Ryff, 1999). While the child sat quietly and read, seven blood pressure readings were taken over a 25-min period with an automated monitor (Critikon Corp., Tampa, FL; Dinamap Model 1846SXP). The average of the final six readings was used, according to standard protocol (Kamarck et al., 1992; Krantz & Falconer, 1995). Overnight (8:00 p.m. to 8:00 a.m.) urine samples were collected, processed, and then deep frozen until subsequent biochemical assays by technicians blind to the child’s cumulative risk status. Epinephrine and norepinephrine were assayed with high-performance liquid chromatography with electrochemical detection (Riggin & Kissinger, 1977), and free cortisol was measured by radioimmune assay (Baxter Travenol Diagnostics, 1987). Creatinine was also assayed in order to control for differences in body mass and incomplete urine voidings (Tietz, 1976).
Table 1 Descriptive Statistics on Cumulative Risk Factors and Developmental Outcomes
Proportion of sample with risk factor
Cumulative risk factors Crowding (no. of people/ room) Noise (Leq, dBA)a Housing problems (0–2) Family separation (0–8) Family turmoil (0–8) Violence (0–5) Income-to-needs ratio Single parent (0/1) Maternal high school dropout (0/1) Cumulative risk (0–9) .61 63.11 .60 1.88 1.65 .18 1.66 .20 7.15 .30 1.38 1.39 .23 1.09 .13 .17 .15 .31 .31 .35 .50 .44 .07 2.39 1.94
Results Cumulative Risk Exposure
Table 1 provides descriptive information on each of the nine risk factors and the socioemotional and physiological stress outcomes. The relatively high proportions of children at risk reflect the fact that half of the sample (50%) consisted of children living at or below the federal poverty line (i.e., income-to-needs ratio 1). Many of the children lived in single-parent households (44%), and approximately a third of the children had been exposed to violence or family turmoil or been separated from their families. Twenty-two percent of the sample had zero risk factors, 18% had one risk, 15% had two risks, 15% had three risks, 11% had four risks, 10% had five risks, and 9% had six or more risks. For the analyses of developmental outcomes, cumulative risks of six or more were combined into one category given the small sample sizes at the upper end. This had no impact on statistical significance. Table 2 depicts the intercorrelations among the outcome measures. Cumulative risk exposure was significantly associated with all of the outcome measures except for diastolic blood pressure. Because there were no statistically significant interaction terms for gender and cumulative risk, all of the analyses were collapsed across gender. Degrees of freedom vary throughout because of missing data.
Developmental outcomes Rutter psychological distress (0–52) Harter global self-worth (1–4) Delayed gratification time (in seconds)b Delayed gratification failure Learned helplessness puzzle persistence (in seconds)c Diastolic blood pressure (mm Hg) Systolic blood pressure (mm Hg) Cortisol ( g/mg creatinine) Epinephrine (ng/mg creatinine) Norepinephrine (ng/mg creatinine) Body mass index (kg/m2) Allostatic load (0–6)d 9.83 3.41 1457.21 6.49 .61 523.86
36% failed to wait 303.41 59.42 102.43 0.029 4.44 32.40 18.33 1.45 175.83 7.10 11.31 0.023 4.17 19.51 3.74 1.22
Psychological Distress and Competence
As can be seen in Table 3, as the number of risk factors rises, maternal reports of psychological distress in the child (Rutter Child Behavior Questionnaire) increase, b 1.13, t(337) 6.61, p .01. Children’s perception of their self-worth (Harter Perceived Competency Scale) was inversely related to cumulative risk, b .06, t(299) 3.28, p .01.
Note. Rutter Rutter Child Behavior Questionnaire; Harter Harter Perceived Competency Scale. a Mean decibel levels of two 2-hr interior recordings on different days. b 1,800 s maximum. c 600 s maximum. d Derived from the total number of biomarkers (resting systolic and diastolic blood pressure; overnight urinary cortisol, epinephrine, and norepinephrine; and body mass index) on which each child fell into the upper quartile of risk.
Self-regulatory behavior, as indicated by delayed gratification, was significantly affected by cumulative risk level, b 74.82, t(199) 3.29, p .01.2 As can be seen in Figure 1, as cumulative risk exposure increased, children were unable to delay gratification for longer periods of time. In addition to measuring the association between cumulative risk and delayed gratification time, I also calculated the association between cumulative risk (i.e., 0 – 6) and whether the child successfully delayed gratification. Successful
delay means the child waited for the full 30 min without signaling (bell), touching the candy, or getting out of his or her chair. Cumulative risk was also significantly related to delayed gratification success, 2(6, N 201) 24.85, p .01.
2 Note that the sample size for the measure of self-regulatory behavior is much smaller because this measure was added to the end of the protocol after the study had begun.
Table 2 Outcome Measures Correlation Matrix
Measure 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.
2 .19** —
3 .16* .01 —
4 .19** .05 .77** —
5 .06 .13* .09 .09 —
6 .04 .01 .17* .12 .11 —
7 .09 .03 .05 .00 .01 .60** —
8 .11 .02 .08 .06 .02 .13* .16* —
9 .06 .04 .02 .01 .16** .09 .05 .09 —
10 .01 .08 .10 .11 .13* .06 .01 .12* .51** —
11 .13* .06 .03 .05 .02 .09 .35** .11 .07 .09 —
12 .01 .04 .09 .02 .07 .46** .52** .38** .49** .45** .37** —
Psychological distress (Rutter) Global self-worth (Harter) Delayed gratification time Delayed gratification successa Helplessness puzzle persistence Diastolic blood pressure Systolic blood pressure Cortisol Epinephrine Norepinephrine Body mass index Allostatic load
Note. Rutter Rutter Child Behavior Questionnaire; Harter 0 successful delay; 1 unable to wait the full 1,800 s. * p .05. ** p .01.
Harter Perceived Competency Scale.
The second behavioral measure of socioemotional adjustment, learned helplessness, was also sensitive to cumulative risk, b 16.10, t(276) 2.37, p .02. As shown in Figure 1, children persisted for a shorter amount of time on the challenging puzzle if they had been exposed to a greater number of multiple risk factors.
the six levels of increasing cumulative risks were 0.96, 1.56, 1.63, 1.90, 1.60, 1.46, and 1.40. As childhood exposure to cumulative risk increased, overall wear and tear on the body was elevated. To examine a possible curvilinear relation, I also tested the quadratic term; it proved significant, b .08, t(247) 3.16, p .01, after controlling for the linear term (J. Cohen & Cohen, 1983).
Figure 2 depicts the blood pressure and cumulative risk data. Systolic, b 1.40, t(271) 3.40, p .01, but not diastolic, b .22, t(271) 1.0, blood pressure reacted to cumulative risk exposure. The significant relations between cumulative risk exposure and the neuroendocrine stress hormones are depicted in Figure 3. All three indicators of neuroendocrine stress activity were elevated by cumulative risk exposure: cortisol, b .002, t(260) 2.70, p .01; epinephrine, b .57, t(263) 3.76, p .01; norepinephrine, b 1.48, t(263) 2.03, p .04. Body mass index (kg/m2) was significantly related to cumulative risk exposure, b 0.36, t(263) 2.47, p .01. Body mass indexes across the six levels of increasing cumulative risks were 17.30, 18.62, 18.23, 19.17, 19.18, 19.11, and 18.46 kg/mg2, respectively. The total allostatic load index was also related to cumulative risk, b .14, t(248) 2.87, p .01. Allostatic load indexes (based on the upper quartiles of six physiological indexes) across
The ecological perspective on human development (Bronfenbrenner & Evans, 2000; Bronfenbrenner & Morris, 1998; Wachs, 2000) posits that development is influenced by the systematic interplay of multiple organismic and environmental (psychosocial, physical) characteristics over time. Traditional statistical interaction terms fail to capture the complexity of such a system of person by environment interaction. An alternative approach based on the concept of cumulative risk exposure has proven fruitful in demonstrating that the natural covariation of multiple risk factors can have potent developmental impacts. In this study, I merged a model of cumulative risk exposure with new developments in neurobiology and physiological stress scholarship (allostasis theory) in order to examine multiple developmental outcomes that may reflect allostatic load on the system. I also extended prior work on cumulative risk and socioemotional development not only by examining subjective reports of well-being but by incorporating two behavioral indicators of socioemotional health, self-regulatory behavior and learned helplessness, into the assessment protocol.
Table 3 Relation of Cumulative Risk to Psychological Distress and Global Self Worth
Number of cumulative risks Variable Rutter psychological distress Harter global self-worth Note. Rutter 0 6.99 3.66 1 8.63 3.43 2 8.96 3.40 3 10.33 3.20 4 12.16 3.21 5 13.55 3.44 6 13.07 3.29
Rutter Child Behavior Questionnaire; Harter
Harter Perceived Competency Scale.
CUMULATIVE RISK AND ALLOSTATIC LOAD
Figure 1. Cumulative risk exposure during early childhood and delay time during delayed gratification (i.e., self-regulatory behavior) and persistence time on a challenging puzzle (i.e., learned helplessness).
Figure 3. Overnight urinary neuroendocrine indices of allostasis (norepinephrine and epinephrine in ng/mg creatinine; cortisol in g/mg creatinine) as a function of cumulative risk exposure.
Moreover, these assessments were made in a population seldom studied, that of rural children. The pattern of results supports the value of such a multimethodological assessment of the developmental correlates of cumulative risk exposure. As indicated in the introduction, prior studies of cumulative risk and children’s adjustment have been restricted to questionnaire measures of psychological distress. Maternal and child reports of psychological distress and competency, respectively, were modestly correlated (r .19). Maternal ratings of
Figure 2. Cardiovascular indices of allostasis as a function of cumulative risk exposure.
distress were correlated with self-regulatory behavior (r .16) and with one index of allostatic load, body mass index (r .13). Children’s own perceptions of self-worth were unrelated to any allostatic load indicators but were positively correlated with task persistence on the learned helplessness puzzle (r .13; see Table 2). Maternal ratings of the child’s psychological distress (see Table 3) are consistent with results from prior studies of cumulative risk and perceived psychological distress among young children (Ackerman et al., 1999; Barocas et al., 1985; Dunst & Trivette, 1994; Fergusson et al., 1994; Furstenburg et al., 1999; Liaw & Brooks-Gunn, 1994; Rutter et al., 1974; Shaw & Emery, 1988; Werner & Smith, 1982). As risk exposure accumulated, children’s psychological distress elevated, up until approximately four risk factors, where it began to level off. For children’s own ratings of competency, there was a dropoff that began at one risk and continued until five accumulated risks, and then there was a slight upturn (see Table 3). The only prior work on cumulative risk and competency (Lengua, 2002) did not include descriptive data but uncovered a significant linear trend for cumulative risk and competency. The present study builds on and extends this prior research on cumulative risk and psychological distress in three ways. First, with the exception of Lengua (2002), prior studies of cumulative risk and socioemotional development have focused on perceived distress. I have shown that both psychological distress and perceived competency are sensitive to cumulative risk exposure. It is worth reiterating that the two subjective socioemotional adjustment measures (maternal ratings of distress and child selfevaluations of self-worth) were only modestly correlated (r .19). I replicated Lengua’s findings of a negative relation between cumulative risk exposure and perceived competency. Second, I have provided the only data on cumulative risk and socioemotional outcomes among a rural population. Prior cumulative risk samples have been restricted to metropolitan areas. Third, I
included information from both the child and the mother. Both sources of information converged in demonstrating adverse socioemotional correlates of early-childhood cumulative risk exposure. The present study supplemented these subjective measures of socioemotional development with multiple methodological assessments of well-being. Both standard behavioral protocols indicative of self-regulatory behavior and learned helplessness, respectively (see Figure 1), as well as cardiovascular (see Figure 2), neuroendocrine (see Figure 3), and fat deposition (body mass index) measures and a total allostatic load index were associated with cumulative risk exposure among young children. To my knowledge, these are the only data on cumulative risk and allostatic load in children; the only data on cumulative risk and susceptibility to learned helplessness; and, except for the study by Barocas et al. (1991), the only results on cumulative risk exposure and selfregulatory behavior. The self-regulatory data are interesting in at least one other respect as well. Metcalf and Mischel (1999) theorized that the development of self-regulatory behavior is a joint function of personal characteristics and experience with chronic adversity. Although there is substantial evidence for links between personal characteristics (e.g., temperament) and self-regulatory behavior, empirical data are lacking on life experiences and the development of self-regulatory behavior (Metcalf & Mischel, 1999; Mischel et al., 1989). As far as I know, the present study and that of Barocas et al. (1991) provide the only empirical evidence of such a connection. Well-documented concurrent and prospective behavioral adjustment problems associated with behavioral dysregulation (Eisenberg, Fabes, & Guthrie, 1997; Mischel et al., 1989) underscore the importance of the association uncovered herein between cumulative risk exposure and delayed gratification. The helplessness and cumulative risk data are consistent with several prior studies showing that singular acute and chronic stressors increase susceptibility to motivational deficits associated with learned helplessness (S. Cohen et al., 1986; Evans, 2001; Peterson et al., 1993). Control-related outcomes are an important and unexplored aspect of cumulative risk exposure. The confluence of multiple risks during early childhood, particularly risks that share to varying degrees the quality of uncontrollability, may threaten feelings of mastery and control and lead to behaviors indicative of learned helplessness (Peterson et al., 1993; Repetti et al., 2002; Taylor, Repetti, & Seeman, 1997). Given the adverse behavioral and academic adjustments associated with low selfefficacy in children and youth (Bandura, 1994; S. Cohen et al., 1986; Peterson et al., 1993; Seligman, 1995), this is an important area for future study. The data also converge with anthropological accounts of the helplessness and despair endemic to children growing up in risky circumstances (Lewis, 1966). Although the absolute levels of physiological and behavioral dysfunction associated with cumulative risk uncovered herein are small, it is important to consider these outcomes in the context of allostasis theory. Recall that allostasis emphasizes the cumulative wear and tear on the body over the life history of the individual as he or she copes with environmental demands. Alterations in basic biological processes set in motion a cascade of altered physiological, cognitive, and behavioral states that portend ill physical and psychological health over time. These alterations in organismic functioning are typically modest when examined in isolation, suggesting few or no health risks. But when considered in concert
with one another, the constellation of small changes in physiological functioning may have major consequences later in life. For example, longitudinal studies of allostatic load among adults replicate long-standing findings that modest changes in individual risk factors (e.g., elevated blood pressure, reduced protective lipids [HDLs]) are associated with small increases in morbidity and mortality. However when these small changes in various physiological markers of risk are considered in total, marked elevations in morbidity and mortality are seen (Karlamangala et al., 2002; Seeman et al., 1997, 2001). Furthermore, these aggregated physiological risk factors are also associated with psychosocial factors such as lower educational attainment, elevated hostility (Kubzansky et al., 1999), and poorer social integration (Seeman et al., 2002). A few investigators have begun to explore allostatic load over the life course. Power, Manor, and Fox (1991) found evidence that accumulated risk factors beginning in childhood (e.g., housing tenure at age 11, behavioral adjustment ratings at age 16) related prospectively to physical health and psychological well-being at age 23 in a large national birth cohort. These same cumulative risks during childhood extended to compromised physical health among 33-year-olds (Power & Matthews, 1998). Felitti and colleagues (1998) revealed that 10 major causes of death in adults were associated with the accumulation of retrospective reports of early childhood experiences of emotional, physical, and sexual abuse plus family dysfunction (e.g., parental substance abuse, domestic violence). Furthermore, exposure to accumulated individual risk factors (i.e., unresponsive parenting in early childhood, low emotional support from significant other during middle adulthood, economic disadvantage in childhood and adulthood, separately) was found to be associated prospectively with higher total allostatic load assessed at age 59 (Singer & Ryff, 1999). An important and challenging measurement question raised by a developmental perspective on allostasis is how to assess early warning signs of elevated allostatic load. What are the appropriate biomarkers to be assessed in young children to reflect early signs of cumulative wear and tear on the body? One tactic is to rely on measures similar to those in adult allostatic load studies that have been shown to be meaningful biomarkers, such as cardiovascular hemodynamics, lipid metabolism, and neuroendocrine activity. These measures, however, primarily reflect what McEwen and Seeman (1999) term secondary outcomes, which are the products of primary biological processes such as cellular events involving local neurotransmitters, trafficking of immune cells, and excitatory amino acids along with glucocorticoid activity. Changes in secondary impacts such as hemodynamics or lipid metabolism may require a longer developmental history to manifest and thus may be more sensitive biomarkers at a later age. Some developmental work also suggests that more dynamic parameters, such as reactivity to an acute demand, may be more sensitive biomarkers in children in relation to chronic stressor exposure than are basal levels of physiological functioning (Allen, Matthews, & Sherman, 1997; Gump, Matthews, & Raikkonen, 1999; Johnston-Brooks, Lewis, Evans, & Whalen, 1998). A related and as yet unanswered question is whether single biomarkers should be used in child studies of exposure to cumulative and single-risk factors or whether multiple biomarkers should be aggregated into total allostatic load metrics, as has been done in the adult literature. Adult studies have operationalized
CUMULATIVE RISK AND ALLOSTATIC LOAD
allostatic load by constructing a summary index across multiple systems. For example, Seeman and colleagues (1997, 2001, 2002) examined 10 biomarkers within the highest quartile of risk (i.e., an allostatic load score of 0 –10 for each participant). This index in turn predicted concurrent and subsequent physical and cognitive pathologies among older adults. To explore the question of the appropriateness of aggregating biomarkers to index allostatic load in young children, I also calculated an aggregated allostatic load index using six individual physiological markers (body mass index, resting diastolic and systolic blood pressure, and overnight epinephrine, norepinephrine, and cortisol). It should be noted that allostatic load metrics in adults have relied on larger sets of physiological measures (typically 10). Cumulative risk exposure in young children was significantly related to allostatic load. Given that a central tenet of allostasis theory is that the accumulation of small changes over time in multiple physiological systems best reflects chronic wear and tear on the body, work with allostatic load in children is probably well served, at least initially, by casting a wide net of physiological measures. As noted in the adult literature on allostasis theory (McEwen, 2000; McEwen & Seeman, 1999), many other viable candidates are also apparent for the assessment of allostatic load, including more dynamic indices such as cardiovascular and neuroendocrine reactivity and recovery in response to acute stressors, plus measures of primary biological activities at the cellular level that appear to underlie many of the more secondary alterations in bodily functioning (e.g., resting blood pressure, overnight urinary neuroendocrine levels, body fat deposition) indicative of allostatic load that have been relied on to date. Another consideration in choosing outcome measures related to cumulative risk and allostatic load during development are tertiary impacts that may be caused by the accumulation of modest shifts in biomarkers across multiple systems (Chen, Matthews, & Boyce, 2002; McEwen, 2000; McEwen & Seeman, 1999; Repetti et al., 2002). For example, HPA-mediated damage in the hippocampus produced by allostatic load (McEwen, 2000) may be related to depression and other symptoms of distress, including low mastery or helplessness. Prefrontal cortical and amygdala atrophy shown in animal models of chronic stress (McEwen, 2000), along with early exposure to violence and unpredictable surroundings, could lead to emotional dysregulation manifested in elevated externalization symptomatology along with more direct indices of poor selfregulatory behavior. Both hypervigilance associated with early experiences of hostility and conflict as well as insufficient recovery periods from these and other highly emotionally arousing states could also lead to dysregulated emotional processes (Repetti et al., 2002). Furthermore, cognitive deterioration, particularly in selective attention as well as in spatial and episodic memory, has been implicated in long-term exposure to elevated glucocorticoids (McEwen, 2000). If the accumulation of exposure to risks heightens allostatic load, eventually one would also expect to see impacts of cumulative risk in psychological distress, especially depression and anxiety, along with lower mastery and increased helplessness, plus signs of emotion dysregulation manifested by hostility, aggression, and poor self-regulatory behavior. Deterioration in selective attention and particular memory processes (spatial and episodic) would be expected as well. Important major questions remain regarding the developmental timing and appropriateness of indices of allostatic load and sub-
sequent symptoms of physical, socioemotional, and cognitive pathology (Chen et al., 2002; McEwen & Seeman, 1999; Repetti et al., 2002; Seeman et al., 2002). Without broad assessments of multiple markers of physiology, plus socioemotional and cognitive outcomes measured over the life course, fundamental questions about how underlying developmental processes are altered by allostatic load cannot be answered. Every individual exposed to high cumulative risk does not manifest elevated allostatic load or other adverse socioemotional, cognitive, or physical outcomes. Research on chronic adversity indicates that warm, responsive parenting may help to offset some of the adverse impacts of early risky environments on physical and psychological morbidity in children (Chen et al., 2002; Repetti et al., 2002; Taylor et al., 1997). Interestingly enough, Seeman et al. (2002) also found that allostatic load was inversely related to social integration in middle-aged and older adults. Several researchers have uncovered factors related to resilience in the face of high cumulative risk during childhood. Among the factors uncovered are intelligence, positive temperament, at least one strong and consistent social bond with an adult, authoritative parenting, and a sense of mastery (Garmezy, 1993; Masten et al., 1999; Rutter 1993; Seifer, Sameroff, Baldwin, & Baldwin, 1992; Werner, 1993; Werner & Smith, 1982). There has also been work examining the conceptual equivalent of cumulative risk, cumulative protection. Two teams of investigators have found evidence, analogous to that from the cumulative risk literature, indicating that resources can also accumulate, providing better protection than single beneficial factors to children with high cumulative risk exposure (Dunst & Trivette, 1994; Zhao, Brooks-Gunn, McLanahan, & Singer, 2000). Although it was not the primary focus of this study, I examined with cluster analysis whether specific patterns of risk factors were associated with allostatic resilience. I attempted to explore the patterns of risk exposure among children with four or more cumulative risks who manifested elevated allostatic load and compared them with those of children at the same high levels of risk who did not manifest high allostatic load. Unfortunately, small sample sizes precluded assessment of this important question. The overall pattern of the data in this study coalesces in a picture of early exposure to cumulative risks positioning children at a young age for potential long-term elevated risk for physical morbidity and behavioral problems (Chen et al., 2002; Repetti et al., 2002; Taylor et al., 1997). Because the pattern of data depends solely on cross-sectional results, long-term follow-up is needed to directly test this hypothesis and to examine the role of exposure duration in cumulative risk and human development. Longitudinal work is needed not only for strengthening causal inference but also because of a central tenet of allostasis theory: The accumulation of risk over time is a critical factor in wear and tear on the body. Trends in contemporary American society, and in Western culture in general, reveal growing chaos in the lives of children, youth, and families (Bronfenbrenner et al., 1996; Garbarino, 1995). Cumulative risk is a valuable heuristic for examining the potential impact of this growing chaos on the health and wellbeing of children and their families. It is important to extend outcome measures of cumulative risk beyond subjective judgments of distress, to incorporate multimethodological indices of development, and to build these assessments into longitudinal designs so that the impacts of early childhood experiences of cumulative risk can be tracked over the life course. Allostasis theory provides a
EVANS Revenson, & J. E. Singer (Eds.), Handbook of health psychology (pp. 365–385). Mahwah, NJ: Erlbaum. Evans, G. W., Hygge, S., & Bullinger, M. (1995). Chronic noise and psychological stress. Psychological Science, 6, 333–338. Evans, G. W., & Kantrowitz, E. (2002). Socioeconomic status and health: The potential role of environmental risk exposure. Annual Review of Public Health, 23, 303–331. Evans, G. W., Wells, N. M., Chan, E., & Saltzman, H. (2000). Housing and mental health. Journal of Consulting and Clinical Psychology, 68, 526 –530. Felitti, V. J., Anda, R., Nordenberg, D., Williamson, D., Spitz, A., Edwards, V., et al. (1998). Relationship of childhood abuse and household dysfunction to many leading causes of death in adults. American Journal of Preventive Medicine, 14, 245–258. Fergusson, D. M., Horwood, J., & Lynskey, M. (1994). The childhoods of multiple problem adolescents: A fifteen year longitudinal study. Journal of Child Psychology and Psychiatry and Allied Disciplines, 35, 1123– 1140. Furstenburg, F. F., Jr., Cook, T. D., Eccles, J., Elder, G. H., Jr., & Sameroff, A. J. (1999). Managing to make it: Urban families and adolescent success. Chicago: University of Chicago Press. Garbarino, J. (1995). Raising children in a socially toxic environment. San Francisco: Jossey-Bass. Garmezy, N. (1993). Children in poverty: Resilience despite risk. Psychiatry, 56, 127–136. Glass, D. C., & Singer, J. E. (1972). Urban stress. New York: Academic Press. Gump, B. B., Matthews, K. A. & Raikkonen, K. (1999). Modeling relationships among socioeconomic status, hostility, cardiovascular reactivity, and left ventricular mass in African American and White children. Health Psychology, 18, 140 –150. Harter, S. (1982). The Perceived Competence Scale for Children. Child Development, 53, 87–97. Hernandez, D. J. (1993). America’s children: Resources from family, government, and the economy. New York: Russell Sage Foundation. Johnston-Brooks, C. H., Lewis, M. A., Evans, G. W., & Whalen, C. K. (1998). Chronic stress and illness in children: The role of allostatic load. Psychosomatic Medicine, 60, 597– 603. Kamarck, T., Jennings, R., Desk, T., Blackman-Weis, E., Johnson, P., Eddy, M., & Manuck, S. (1992). Reliable measures of behaviorally evoked cardiovascular reactivity from a PC based test battery. Psychophysiology, 29, 17–28. Karlamangala, A. S., Singer, B. H., McEwen, B. S., Rowe, J. W., & Seeman, T. E. (2002). Allostatic load as a predictor of functional decline: MacArthur studies of successful aging. Journal of Clinical Epidemiology, 55, 696 –710. Krantz, D. S., & Falconer, J. (1995). Measurement of cardiovascular responses. In S. Cohen, R. C. Kessler, & L. Underwood (Eds.), Measuring stress (pp. 193–212). New York: Oxford University Press. Kubzansky, L. D., Kawachi, I., & Sparrow, D. (1999). Socioeconomic status, hostility, and risk factor clustering in the normative aging study: Any help from the concept of allostatic load? Annals of Behavioral Medicine, 21, 330 –338. Lengua, L. J. (2002). The contribution of emotionality and self-regulation to the understanding of children’s response to multiple risk. Child Development, 73, 144 –161. Lepore, S. J., & Evans, G. W. (1996). Coping with multiple stressors in the environment. In M. Zeidner & N. Endler (Eds.), Handbook of coping (pp. 350 –377). New York: Wiley. Lewis, O. (1966). The culture of poverty. Scientific American, 215, 19 –25. Liaw, F., & Brooks-Gunn, J. (1994). Cumulative familial risk and low birth weight children’s cognitive and behavioral development. Journal of Clinical Child Psychology, 23, 360 –372. Masten, A. S., Hubbard, J., Gest, S. D., Tellegen, A., Garmezy, N., &
new and potentially important perspective on cumulative risk and human development. It offers a viable set of physiological processes marking chronic wear and tear on the body that may help explain how cumulative risk exposure over the life course can lead to initially small, subtle changes in organismic functioning during childhood that can accumulate over time into serious physical and psychological morbidity later in life.
Ackerman, B. P., Izard, C., Schoff, K., Youngstrom, E., & Kogos, J. (1999). Contextual risk, caregiver emotionality, and the problem behaviors of six and seven year old children from economically disadvantaged families. Child Development, 70, 1415–1427. Allen, M. T., Matthews, K. A., & Sherman, F. S. (1997). Cardiovascular reactivity to stress and left ventricular mass in youth. Hypertension, 30, 782–787. Bandura, A. (1994). Self-efficacy: The exercise of control. San Francisco: W. H. Freeman. Barocas, R., Seifer, R., & Sameroff, A. J. (1985). Defining environmental risk: Multiple dimensions of psychological vulnerability. American Journal of Community Psychology, 13, 433– 447. Barocas, R., Seifer, R., Sameroff, A. J., Andrews, T. A., Croft, R., & Ostrow, E. (1991). Social and interpersonal determinants of developmental risk. Developmental Psychology, 27, 479 – 488. Baxter Travenol Diagnostics. (1987). Gammacoat I125 cortisol radioimmune assay. Cambridge, MA: Author. Boyle, M., & Jones, S. (1985). Selecting measures of emotional and behavioral disorders of childhood for use in general populations. Journal of Clinical Psychology and Psychiatry, 26, 137–159. Bronfenbrenner, U., & Evans, G. W. (2000). Developmental science in the 21st century: Emerging theoretical models, research designs, and empirical findings. Social Development, 9, 115–125. Bronfenbrenner, U., McClelland, P., Wethington, E., Moen, P., & Ceci, S. (1996). The state of Americans: This generation and the next. New York: Free Press. Bronfenbrenner, U., & Morris, P. (1998). The ecology of developmental process. In W. Damon & R. Lerner (Eds.), Handbook of child psychology (5th ed., pp. 992–1028). New York: Wiley. Bullinger, M., Hygge, S., Evans, G. W., Meis, M., & van Mackensen, S. (1999). The psychological cost of airport noise for children. Zentralblatt fur Hygiene und Umweltmedizin, 202, 197–213. Chen, E., Matthews, K. A., & Boyce, T. (2002). Socioeconomic differences in health: What are the implications for children? Psychological Bulletin, 128, 295–329. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Cohen, S. (1980). Aftereffects of stress on human performance and social behavior: A review of research and theory. Psychological Bulletin, 88, 82–108. Cohen, S., Evans, G. W., Stokols, D., & Krantz, D. S. (1986). Behavior, health, and environmental stress. New York: Plenum Press. Dawes, R. M., & Corrigan, B. (1974). Linear models in decision making. Psychological Bulletin, 61, 95–106. Dunst, C. J., & Trivette, C. M. (1994). Methodological considerations and strategies for studying the long term effects of early interventions. In S. L. Friedman & H. C. Haywood (Eds.), Developmental follow-up (pp. 277–313). New York: Academic Press. Eisenberg, N., Fabes, R. & Guthrie, I. (1997). Coping with stress: The roles of regulation and development. In I. Sandler & S. A. Wolchik (Eds.), Handbook of children’s coping with common stressors (pp. 41–70). New York: Plenum Press. Evans, G. W. (2001). Environmental stress and health. In A. Baum, T.
CUMULATIVE RISK AND ALLOSTATIC LOAD Ramirez, M. (1999). Competence in the context of adversity: Pathways to resilience and maladaptation from childhood into late adolescence. Development and Psychopathology, 11, 143–169. McEwen, B. S. (1998). Protective and damaging effects of stress mediators. New England Journal of Medicine, 338, 171–179. McEwen, B. S. (2000). The neurobiology of stress: From serendipity to clinical relevance. Brain Research, 886, 172–189. McEwen, B. S., & Seeman, T. E. (1999). Protective and damaging effects of mediators of stress: Elaborating and testing the concepts of allostasis and allostatic load. In N. E. Adler, M. Marmot, B. S. McEwen, & J. Stewart (Eds.), Socioeconomic status and health in industrial nations (pp. 30 – 47). New York: New York Academy of Sciences. McEwen, B. S., & Stellar, E. (1993). Stress and the individual: Mechanisms leading to disease. Annals of Internal Medicine, 153, 2093–2101. Metcalf, J., & Mischel, W. (1999). A hot/cold system analysis of delay of gratification: Dynamics of willpower. Psychological Review, 106, 3–19. Mischel, W., Shoda, Y., & Rodriguez, M. (1989). Delay of gratification in children. Science, 244, 933–938. Peterson, C., Maier, S., & Seligman, M. E. P. (1993). Learned helplessness. New York: Oxford University Press. Power, C., Manor, O., & Fox, J. (1991). Health and class. London: Chapman & Hall. Power, C., & Matthews, S. (1998). Accumulation of health risks across social groups in a national longitudinal study. In S. S. Strickland & P. S. Shetty (Eds.), Human biology and social inequality (pp. 36 –57). New York: Cambridge University Press. Repetti, R. L., Taylor, S. E., & Seeman, T. E. (2002). Risky families: Family social environments and the mental and physical health of offspring. Psychological Bulletin, 128, 330 –366. Riggin, R., & Kissinger, P. (1977). Determination of catecholamines in urine by reverse phase liquid chromatography with electrochemical detection. Analytic Chemistry, 49, 2109 –2111. Rutter, M. (1983). Statistical and personal interactions: Facets and perspectives. In D. Magnusson & V. Allen (Eds.), Human development: An interactional perspective (pp. 295–319). New York: Academic Press. Rutter, M. (1993). Stress, coping, and development. In N. Garmezy & M. Rutter (Eds.), Stress, coping, and development in children (pp. 1– 41). New York: McGraw-Hill. Rutter, M., Tizzard, J., & Whitmore, K. (1970). Education, health, and behavior. London: Longmans. Rutter, M., Yule, B., Quinton, D., Rowlands, O., Yule, W., & Berger, M. (1974). Attainment and adjustment in two geographic areas: III. Some factors accounting for area differences. British Journal of Psychiatry, 125, 520 –533. Sameroff, A. J. (1998). Environmental risk factors in infancy. Pediatrics, 102, 1287–1292. Schell, L. M. (1997). Culture as a stressor: A revised model of biocultural interaction. American Journal of Physical Anthropology, 102, 67–77. Seeman, T. E., & McEwen, B. S. (1996). Impact of social environmental characteristics on neuroendocrine regulation. Psychosomatic Medicine, 58, 459 – 471. Seeman, T. E., McEwen, B. S., Rowe, J. W., & Singer, B. H. (2001). Allostatic load as a marker of cumulative biological risk: MacArthur studies of successful aging. Proceedings of the National Academy of Sciences, 98, 4770 – 4775. Seeman, T. E., Singer, B. H., Rowe, J. W., Horwitz, R., & McEwen, B. S. (1997). Price of adaptation—allostatic load and its health consequences:
MacArthur studies of successful aging. Archives of Internal Medicine, 157, 2259 –2268. Seeman, T. E., Singer, B. H., Ryff, C. D., Love, G. D., & Levy-Storms, L. (2002). Social relationships, gender, and allostatic load across two age cohorts. Psychosomatic Medicine, 64, 395– 406. Seifer, R., Sameroff, A. J., Baldwin, C., & Baldwin, A. (1992). Child and family factors that ameliorate risk between 4 and 13 years of age. Journal of the American Academy of Child and Adolescent Psychiatry, 31, 893–903. Seligman, M. E. P. (1975). Learned helplessness. San Francisco: W. H. Freeman. Seligman, M. E. P. (1995). The optimistic child. New York: HoughtonMifflin. Shaw, D. S., & Emery, R. E. (1988). Chronic family adversity and school-age children’s adjustment. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 200 –206. Sherman, A. (1992). Falling by the wayside: Children in rural America. Washington, DC: Children’s Defense Fund. Singer, B. H., & Ryff, C. D. (1999). Hierarchies of life histories and associated health risks. In N. E. Adler, M. Marmot, B. S. McEwen, & J. Stewart (Eds.), Socioeconomic status and health in industrial nations (pp. 96 –115). New York: New York Academy of Sciences. Sterling, P., & Eyer, J. (1988). Allostasis: A new paradigm to explain arousal pathology. In S. Fisher & J. Reason (Eds.), Handbook of life stress, cognition, and health (pp. 629 – 649). New York: Wiley. Taylor, S. E., Repetti, R. L., & Seeman, T. E. (1997). Health psychology: What is an unhealthy environment and how does it get under the skin? Annual Review of Psychology, 48, 411– 447. Tietz, N. (1976). Fundamentals of clinical chemistry (2nd ed.). Philadelphia: W. B. Saunders. Wachs, T. D. (2000). Meaningful differences. Washington, DC: American Psychological Association. Werner, E. E. (1993). Risk, resilience, and recovery: Perspectives from the Kauai longitudinal study. Development and Psychopathology, 5, 503– 515. Werner, E. E., & Smith, R. S. (1982). Vulnerable but invincible. New York: McGraw-Hill. Williams, S., Anderson, J., McGee, R., & Silva, P. (1990). Risk factors for behavioral and emotional disorder in preadolescent children. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 413– 419. Work, W., Cowen, E., Parker, G., & Wyman, P. (1990). Stress resilient children in an urban setting. Journal of Primary Prevention, 11, 3–19. Wyman, P., Cowen, E., Work, W., & Parker, G. (1991). Developmental and family milieu correlates of resilience in urban children who have experienced major life stress. American Journal of Community Psychology, 19, 405– 426. Zhao, H., Brooks-Gunn, J., McLanahan, S., & Singer, B. H. (2000). Study the real child rather than the ideal child: Bringing the person into developmental studies. In L. Bergman, R. B. Cairns, L. G. Nilsson, & L. Nystedt (Eds.), Developmental science and the holistic approach (pp. 393– 419). Mahwah, NJ: Erlbaum.
Received June 11, 2002 Revision received December 27, 2002 Accepted December 31, 2002