Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Feb 27.
Published in final edited form as: J Aging Health. 2010 May 28;22(6):748–771. doi: 10.1177/0898264310367516

Stressors over the life course and neuroendocrine system dysregulation in Costa Rica

Omer Gersten 1, William H Dow 2, Luis Rosero-Bixby 3
PMCID: PMC3937264  NIHMSID: NIHMS553299  PMID: 20511582

Abstract

Objectives

A key aspect of the increasingly popular allostatic load (AL) framework is that stressors experienced over the entire life course result in physiological dysregulation. Although core to AL theory, this idea has been little tested and where it has been tested the results have been mixed.

Methods

We analyze the CRELES, a new, cross-sectional, and nationally representative survey of older Costa Rican men and women (ages 60-109). The survey was conducted in 2004-2006 and has a sample size of 2,827. This paper focuses on the relation between a variety of stressors experienced over the life course and cortisol, dehydroepiandrosterone sulfate (DHEAS), epinephrine, and norepinephrine analyzed separately and in an index.

Results

Some links were found between certain stressors and worse cortisol levels, but overall almost all of the stressors examined were not associated with riskier neuroendocrine biomarker profiles.

Discussion

More work is needed to attempt to establish the connection between stressors experienced over the life course and resting levels of the neuroendocrine markers.

Keywords: stress, biomarkers, neuroendocrine allostatic load (NAL), Costa Rica, aging

Introduction

At least two important testable hypotheses stem from the allostatic framework, a framework that has grown markedly in popularity and has emphasized the role of stress in illness (Gersten, 2008a; McEwen, 2004). One hypothesis is that allostatic load (AL), a measure of physiological dysregulation, is the result, over extended periods of time, of repeated activation of the body’s adaptive processes in response to challenge. Another hypothesis is that AL is a risk factor for morbidity and mortality. Of these two hypotheses, far more support has been found for the latter. Using the MacArthur studies, for instance, Seeman and others (1997) found that high AL at baseline predicted greater cognitive and physical declines and earlier mortality over the study period. In addition, work by Goldman and others (2006) and Turra and others (2005) found that various measures of physiological dysregulation in a Taiwanese population predicted health outcomes such as depression, cognitive and physical function, and survival.

In contrast to these findings, using the same Taiwanese data set, analysis by Glei and others (2007) attempting to link various stressors to riskier AL levels produced mixed results. Although an index of stressors with items such as moving, crime victimization, and the death of a child experienced between 1996 and 2000 was associated (although quite weakly) with higher AL levels, other stressors such as low socioeconomic status, low levels of emotional support, and fewer social ties with friends and relatives were not. Gersten (2008b) also analyzed the same data set as Glei and others, but with a focus on neuroendocrine markers and a different set of stressors over the (e.g., length of widowhood and subjective reports of chronic stress over family matters and work and financial issues), and found almost entirely negative results. And although negative findings have also been found by others (Babisch et al., 2001; Powell et al., 2002), these studies typically have only examined one particular type of stressor (e.g., marital disruption or noise exposure) instead of multiple stressors that span the life course. A further limitation of these studies is that they have often investigated one particular biomarker (instead of a grouping representative of a physiological system or systems) and have been conducted on small, non-population-based samples (Babisch et al., 2001; Luecken et al., 1997; Powell et al., 2002; Wheler et al., 2006). In sum, there is a paucity of studies (the 2000 SEBAS and perhaps one or two others) that can claim to have tested in a rigorous way the hypothesis that markers of life history stress are correlated with higher levels of AL. In order to extend this literature, then, we will analyze a new, nationally representative data set from Costa Rica.

The data that will be analyzed in this paper comes from the CRELES, which recently obtained information from older Costa Rican men and women. Much of the data was meant to be comparable to other studies that have investigated AL, such as the MacArthur studies and the Taiwanese SEBAS, and thus the CRELES has obtained many of the same biomarkers as these surveys. One drawback of the CRELES is the lack of questions that probe subjective levels of stress (e.g., “Do you feel stressed about…”). A strength of the survey, however, is its collection of a number of indicators of stressful life events, especially those occurring in early childhood (e.g., economic deprivation). Many other surveys investigating the impact of life stressors on AL have only examined stressors that have occurred in middle and later life (Goldman et al., 2005; Seeman et al., 2004), even though the allostatic framework is quite clear about the importance of using a life course approach in analyses (Crimmins & Seeman, 2004; McEwen, 2004).

As suggested earlier, AL is the idea that the body experiences a cost, or “wear and tear,” from responding to myriad acute and chronic challenges/stressors over the life course (McEwen, 1988; Timiras & Gersten, 2007). AL is also thought to develop in a number of different and important physiological systems, including the metabolic, cardiovascular, and neuroendocrine ones (McEwen, 1988; Timiras & Gersten, 2007). The paper here will focus on the neuroendocrine markers of the AL construct for a number of reasons. First, in population-level studies that have been conducted to date, the neuroendocrine markers have been some of the most recently added and hence least studied (compared to, say, those markers indicative of cardiovascular and metabolic function). Biomarkers of neuroendocrine system function have been little studied even though they are critical to the stress response and form a core component of the AL measure. Second, despite the recent inclusion of neuroendocrine markers in large-scale studies, there is convincing evidence that certain levels of the markers predict a number of health problems, including more rapid decline in physical and cognitive function, greater incidence of cardiovascular disease, and earlier mortality (Goldman et al., 2006; Karlamangla et al., 2005; Seeman et al., 2001). In other words, the neuroendocrine markers make an important contribution in predicting worse health. Third, although one of the strengths of the AL construct has been measurement of different physiological systems in one index in an attempt to gauge health more holistically, such an approach is also one of the construct’s weaknesses. That is, from a physiological perspective, it can be difficult to interpret a score from the measure that includes such vastly different markers. Relatedly, it is often unclear which system, if any, is driving an overall pattern of the construct. Fourth, a focus on the neuroendocrine markers in this paper allows for a more in-depth analysis of them than can usually be carried out. Such a careful treatment is important for a number of reasons including that, as mentioned before, these markers are relatively novel and relatively little is known about them. Thus, for the above reasons, this paper will focus on analyzing four biomarkers representing neuroendocrine system function (i.e., cortisol, dehydroepiandrosterone sulfate (DHEAS), epinephrine, and norepinephrine) in relation to stressors experienced over life course.

Study Hypotheses

Based on the general literature and that specific to Costa Rica, and following in the tradition of the “environmental stress perspective” that focuses on potentially stressful life events (Cohen, Kessler, & Gordon, 1995), we hypothesize that a number of states and experiences have likely proved challenging and hence have likely led to greater AL. Since life brings both good and bad, growing older can only result in exposure to more stressors. And because AL is thought to be cumulative, we hypothesize that greater age is positively correlated with greater AL. We also expect a similar relationship between female sex and AL since women generally report greater distress and depression than men (Steptoe et al., 2007; Thoits, 1995). Indicators of lesser material resources, such as lower education, lower current household wealth, and economic deprivation early in life are all expected to be associated with higher AL (Dow & Schmeer, 2003; Rosero-Bixby, Dow, & Lacle, 2005; Steptoe et al., 2007; Wolf, De Andraca, & Lozoff, 2002), and so too are indicators of lesser emotional resources, such as growing up without a biological father (Budowski & Rosero-Bixby, 2003) and earlier maternal age at death. Markers of poor health early in life, like having had malaria or asthma, might also very well indicate greater stress (and hence greater AL) that comes with dealing with illness.

Given the suggestion in the literature that some of the negative health effects of social deprivation are due to increased levels of stress (Cacioppo & Hawkley, 2003), we expect that measures of such deprivation (which in our study are whether the respondents are unmarried, live alone, and attend church infrequently or not at all) should also be associated with higher AL (Brenes-Camacho, 2008; Dow & Schmeer, 2003; Low, 1981; Rosero-Bixby, Dow, & Lacle, 2005; Wolf, De Andraca, & Lozoff, 2002). Further, we expect a similar relationship between AL and measures of personal loss, which in this study are the experience of a death of a child and length of widowhood (Low, 1981). Status as an immigrant (in comparison to the native born) is included as a variable in our analysis since such status is potentially important, though we remain neutral in hypothesizing its directionality. On the one hand, immigrants are more likely to be disconnected from family, experience more difficult working and living situations, and experience discrimination (Bolaños et al., 2008; Sandoval-García, 2004), but on the other hand they may be healthier and more robust to these sorts of stressors than the native population because of selective migration (Herring et al., in press). Moreover, we suspect that rural residents of Costa Rica, who compare less favorably than their urban counterparts on a number of indicators of welfare (e.g., employment rate, infant mortality, and levels of malnutrition (Bähr & Wehrhahn, 1993; Hall, 1984)), will have higher ALs. Lastly, in analysis of only the currently married, we predict that those who report a spouse in poor health will themselves have higher AL in part because of the possibility of stress caused by caregiving responsibilities (Epel et al., 2004).

Data and Methods

Overview of the data set

We analyze the Costa Rican Study on Longevity and Healthy Aging (CRELES), a population survey conducted in Costa Rica in 2004-2006 (for a more detailed description of the study consult Rosero-Bixby (2007)). The survey is nationally representative of those 60 and older in the non-institutionalized population and the CRELES drew its sub-sample of respondents from the 2000 census database. Among other things, the interview portion of the CRELES included questions about cognitive and physical functioning, health care utilization, nutrition and other health behaviors, social support, employment history and pensions, and a variety of life stressors. The in-home interviews averaged nearly an hour and a half and during the same visit by survey staff mobility tests were performed and blood pressure measurements were taken. With the respondents’ additional consent, they were enrolled in the more invasive aspect of the survey’s data collection efforts. After receiving relevant instructions and materials, participants collected urine and began fasting on the same day as the in-home interview and on the next day the survey staff picked up the urine, drew blood samples, and took anthropometric (e.g., height and weight) measures. The blood and urine samples were used to determine traditional health indicators such as total and HDL cholesterol and less traditional indicators such as epinephrine and cortisol.

Of survivors who could be located and were initially contacted for inclusion in the 2004-2006 CRELES, 96% gave interviews, yielding a sample of 2,827 participants. Of these, 95 and 92% gave blood and urine samples, respectively, and in about 25% of all cases a proxy (most often the respondent’s son or daughter) helped answer some questions for the respondent. The survey over-sampled those over 95 years old.

Dependent variable

The neuroendocrine biomarkers

In this paper we focus on cortisol, DHEAS, epinephrine, and norepinephrine, a physiologically coherent class of markers indicative of neuroendocrine system function (Cohen, Kessler, & Gordon, 1995; Crimmins & Seeman, 2001; Sapolsky, 2004). The measure used here based on these markers is called NAL, for neuroendocrine allostatic load, and has been discussed in more detail elsewhere (Gersten, 2008b). Among NAL’s greatest advantages is its interpretability that stems from grouping markers of a single physiological system. NAL includes markers related to two neuroendocrine systems: the hypothalamic-pituitary-adrenal (HPA) axis and the sympathetic nervous system (SNS). The HPA axis is key in regulating homeostatic processes in the body and environmental stressors can lead it as well other regulatory systems to react (Cohen, Kessler, & Gordon, 1995; Crimmins & Seeman, 2001; Sapolsky, 2004). Cortisol and DHEAS are indicators of HPA axis activity. The body’s “fight or flight” response is in part mobilized by the SNS and its activity can be measured by norepinephrine and epinephrine levels (Cohen, Kessler, & Gordon, 1995; Crimmins & Seeman, 2001; Sapolsky, 2004).

Measurement of biomarkers

The CRELES attempted to capture basal levels of the neuroendocrine biomarkers and to this end the blood and urine samples were collected in the participants’ homes under resting conditions. Three of the four markers were collected in overnight urine samples and when collected in this way the samples represent integrated, in contrast to point-in-time, measures. For cortisol, norepinephrine, and epinephrine, respondents were asked to void urine at 6pm, which was discarded, and to collect all subsequent samples until 6am the following day. In part because dissimilar body size leads to differential concentration of the neuroendocrine markers in the urine, total urine was standardized using grams of creatinine. The subjects also fasted from 6 p.m. onwards on the day they began urine collection until a study affiliate came to their home to collect the urine sample and draw blood. The amount of DHEAS in the body was determined through these blood samples.

The blood samples for each respondent were drawn by venipuncture by a phlebotomist and put into three tubes, one tube with EDTA (which acts as an anticoagulant) and two serum-separating tubes (SST) with clot activators. The tubes with the clot activator were centrifuged in the field to separate the serum from the other elements and to prevent hemolysis (the breaking open of red blood cells and the release of hemoglobin into the surrounding fluid). From the point of initial blood collection, the tubes were kept in coolers for no more than six hours until they were separated in various nearby labs into aliquots of 0.5 or 1.0 mL and stored at −40°C. The serum was later used to analyze DHEAS levels and levels of some other markers. Regarding the urine samples, after they were picked up from respondents’ homes they were kept in coolers and also taken to nearby labs where their volume was measured and separated into five aliquots (of either 1.0 or 2.0 mL). These aliquots of urine were also stored at −40°C.

From the storage labs, the urine samples were initially sent to Neuroscience Laboratories at the University of Costa Rica for analysis and then to the Central American Center for the Analysis of Hormones (CENHACE), a private laboratory in San Jose, Costa Rica. Both these laboratories were certified by a national reference center of clinical chemistry, an agency under the Ministry of Health. The catecholamines were analyzed by high performance liquid chromatography (HPLC) at Neuroscience Laboratories and cortisol and DHEAS were analyzed by chemiluminescence immunoassay at the CENHACE laboratory. Unfortunately, for a sizable share of respondents (about 37%) that otherwise gave valid urine samples, epinephrine and norepinephrine values had to be coded as missing since there was a problem with acidifying the urine samples properly. As far as the possibility of biased esitmates from the missing values on eipinephrine and norepinephrine is concerned, the missingness is related to when respondents were surveyed during the survey cycle (i.e., between November 2004 and September 2006) which is unrelated to individual characteristics and thus there is no reason to suspect that the missingness is systematic1. It is noteworthy that no such missingness issue applies to cortisol and DHEAS for which the CRELES has quite a large sample. Also, the sample size of the CRELES for which there is complete data on all four neuroendocrine markers (that is, including epinephrine and norepinephrine) is still larger than other similar, key studies2.

Independent variables

Most of the independent variables used are straightforward to interpret, although the following require some explanation. Household wealth is determined by first creating an index based on whether the respondent’s home has a kitchen, electricity or gas as cooking fuel, potable water, indoor toilet, a refrigerator or freezer, television, a phone (either a cell or landline), and washing machine, and whether the respondent owns a car. This index is then coded into high, medium, and low household wealth categories. Economic problems early in life were determined by asking respondents whether or not in childhood and adolescence they lived in a home that had a bathroom or latrine, lived in a home that had electricity, slept on the floor or with others in a bed, and regularly wore shoes. Health problems early in life were determined by whether or not the respondent reported having in childhood and adolescence tuberculosis, rheumatic fever, poliomyelitis, malaria, and asthma/chronic bronchitis. It is worth noting here a limitation of the recall variables, and that is the possibility of negatively biased recall of childhood variables in depressed and otherwise unhealthy participants who are also likely to have abnormal biomarker levels.

In another finance-related question, respondents were asked to describe their present economic situation, to which they could respond “excellent,” “very good,” “good,” “average/normal,” or “bad.” Respondents were also asked to provide their total monthly income stemming from work and pensions. Respondents who did not give a precise figure, but gave a range (e.g., 80,000-170,000 colones/month) were given the mean income of those reporting an exact amount within the same range. Income from pensions and work were added to that from transfers from relatives and friends to produce the final variable of total monthly income.

In drawing on an established literature of stressful life events checklists (Turner & Wheaton, 1995), we created a measure of cumulative adversity in which respondents received one point toward their score if they could be characterized by any of the following: less than six years of education, rural residence, lower household wealth, “bad” self-assessed economic situation, monthly income less than or equal to the lower 25th percentile of incomes, being currently unmarried, living alone, death of at least one child, less than weekly religious attendance, mother with no formal education, mother who died before fifty years old, growing up without a biological father, having reported one or more health problems early in life, and having reported three or more economic problems early in life. This cumulative adversity index can thus theoretically range from 0-14. It should be noted that limitations of the index include that the variable growing up without a biological father in the household does not make clear whether the respondent was raised by a single mom alone or with the additional help of another father figure. Also, having information on when participants’ mothers died is less preferable to knowing exactly how old participants themselves were when their mothers died. Lastly, in addition to the cumulative adversity index, we created indices that are subsets of the larger measure and based on the categories of socioeconomic status, social deprivation, loss, and early childhood conditions.

Since levels of the neuroendocrine biomarkers can be influenced by a variety of factors independent of stress (Cohen, Kessler, & Gordon, 1995; James & Brown, 1997), all models initially controlled for smoking, alcohol consumption, and medication use. We present results without these controls, however, since simpler models without the controls are nearly identical to those with them.

Methods

Regarding extreme values, one outlier was removed for norepinephrine, two for epinephrine, and four for cortisol. These outliers were at least 11 standard deviations above the mean for their respective distributions. Concerning other data transformations, the four neuroendocrine biomarkers all had distributions that exhibited positive skewness (right tails) and have been logged, creating more normalized distributions and more normalized residuals.

In this paper we will analyze the biomarkers both individually and as part of an index. When analyzed individually, the biomarkers are kept continuous and are the dependent variables in separate OLS regressions. When analyzed in an index, we follow the most popular approach to operationalizing AL which is to create a score that gives one point for every biomarker for which the subject can be considered at higher risk (i.e., the elevated risk zone approach) (Glei et al., 2007; Seeman et al., 1997; Singer & Ryff, 1999). The literature most often represents high risk by greater values for cortisol, epinephrine, and norepinephrine, and lower values for DHEAS (Kubzansky, Kawachi, & Sparrow, 1999; Loucks, Juster, & Pruessner, 2008; Seeman et al., 1997); this convention is followed here. Since there is no agreed upon standard for what biomarker values represent different risk levels, it has been most common to define risk as above or below distribution percentiles (e.g., the 10th, 25th, 75th, and 90th) (Kubzansky, Kawachi, & Sparrow, 1999; Goldman et al., 2005; Seeman et al., 1997). See Table 1 for descriptive statistics and cut-points for the neuroendocrine biomarkers. Since subjects can be assigned 1 point on each of the four biomarkers if they have high risk values, NAL scores can range from 0–4. These scores serve as the dependent variables in OLS regressions (using Poisson regressions instead produces only minor differences in the results (not shown)). Lastly, all analysis is carried out using STATA version 9.0 (StataCorp, 2005) and the multivariate analysis makes use of sample weights which correct for the oversampling of the oldest-old and for some non-response by demographics.

Table 1. Descriptive statistics and cut-points for the neuroendocrine biomarkers – sample population, Costa Rica (ages 60 to 109, both sexes combined, years 2004-2006)a.

Percentile cutoffs
Mean SD N 10th 25th 75th 90th
System
 HPA axis
  Cortisol (logged)b 3.09 0.72 2252 2.21 2.62 3.53 3.96
  DHEAS (logged)c 3.58 0.79 2621 2.34 2.98 -- --
 SNS
  Epinephrine (logged)b 1.62 1.00 1520 -- -- 2.32 2.94
  Norepinephrine (logged)b 3.52 0.69 1571 -- -- 3.96 4.36
Open in a new tab
a

The tabulations are based on unweighted survey data.

b

μg/g creatinine.

c

μg/dl.

Source: Authors’ tabulations based on the 2004-2006 CRELES (Rosero-Bixby, 2007).

Results

Table 2 depicts descriptive statistics for variables that are used in this analysis. One of the things to note in the table is the relatively low levels of education for these Costa Ricans, with 51% not having completed their primary education (i.e., having less than six years of schooling). Also striking is the percent who have had at least one of their children die and the percent who have grown up without a biological father (36% and 21%, respectively). Table 2 also reveals that religion is important in the lives of many older persons in Costa Rica, as nearly 50% reported going to church one or more times a week. Lastly, it is also worth observing that 36% of those with a spouse reported that the spouse has a serious health problem, suggesting that a fair amount of married older persons provide caregiving services to their husband or wife.

Table 2. Descriptive statistics for the dependent and independent variables used in the analysis – sample population, Costa Rica (ages 60 to 109, both sexes combined, years 2004-2006)a.

Variables % or Mean (SD) Range N
Dependent
 Neuroendocrine allostatic load (NAL)b 0.85 (0.96) 0-4 1332
Independent
 Demographic
  Age 70.5 (8.11) 60-109 2827
  Female sex 53% -- 2827
  Low education (< 6 years) 51% -- 2827
  Rural residence (v. urban) 37% -- 2827
  Immigrant (v. native born) 5% -- 2817
 Economic resources
  Household wealthc 2.13 (0.62) 1-3 2780
  Monthly income (colones in thousands) 190.23 (539.2) 0-10,548 2738
  Self-assessed economic situationd 3.6 (0.94) 1-5 2811
 Spousal characteristics
  Low education (< 6 years) 43% -- 2827
  Serious health problem 36% -- 1402
 Social deprivation
  Currently unmarried (v. curr. married) 40% -- 2817
  Lives alone 10% -- 2823
  Low church attendance (< weekly) 50% -- 2822
 Loss
  No. of children who have died (>= 1) 36% -- 2818
  Length of widowhood (years)e 14.1 (12.0) 0-70 785
 Early childhood conditions
  Maternal age at death 74.8 (17.1) 17-115 2302
  Low maternal education (no edu.) 69% -- 2245
  Lived without biological father 21% -- 2114
  Poor health (>= 1 health problems) 22% -- 2090
  Economic deprivation indexf 2.0 (1.3) 0-4 2103
 Cumulative adversity
  Overall stressor index 3.8 (2.3) 0-13 1764
Open in a new tab

Note:

a

Tabulations based on weighted data.

b

Respondents received one point toward their neuroendocrine allostatic load (NAL) score for each biomarker which had a “high-risk” value (i.e., a value below the 25th or above the 75th percentiles).

c

High wealth is coded as three and low wealth is coded as one.

d

“Excellent” is coded as one and “bad” as five.

e

Only includes the widowed respondents.

f

More severe economic deprivation is represented by higher values on this index.

Source: Authors’ tabulations based on the 2004-2006 CRELES (Rosero-Bixby, 2007.

Table 3 presents results from regressions in which each stressor variable is included in a separate regression with age and sex as the only controls. With regressions for age and sex also run (controlling for sex and age, respectively), each dependent variable in the table is associated with 17 different regressions. As can be seen from the table, the strongest and most consistent relationship is that for age and sex. Being female is associated with riskier levels for all the biomarkers, with the relationship most statistically significant for DHEAS and norepinephrine. Greater age is also associated with worse biomarker profiles for all biomarkers. For the stressor variables, the relationships are, by and large, far from statistically significant at the conventional threshold, inconsistent, or going in the unexpected direction. For example, as expected, low education is associated with worse biomarker profiles for cortisol and norepinephrine, but associated with a good profile for DHEAS. A similar relationship holds for rural residence. Although for cortisol, as expected, greater household wealth is linked to more favorable cortisol values, poor health (unexpectedly) is also linked to more favorable cortisol values. As regards lack of associations, none of the stressor variables are statistically significant for epinephrine and the social deprivation and loss variables (and most of the early childhood conditions variables) are not statistically significant for the other biomarkers as well.

Table 3. Estimated regression results from separate regressions for each of the stressor variables and each biomarker, controlling only for age and sex, Costa Rica (ages 60 to 109, both sexes combined, years 2004-2006)a.

Dependent variablesb
Independent variables Cortisol DHEAS Epinephrine Norepinephrine
Demographic
 Age 0.01 (0.002)*** −0.02 (0.002)*** 0.01 (0.037)* 0.01 (0.002 )*
 Female sex 0.09 (0.039)* −0.62 (0.033)*** 0.16 (0.062)* 0.28 (0.040)***
 Low education (< 6 years) 0.11 (0.038)** 0.15 (0.033)*** 0.05 (0.064) 0.07 (0.040)+
 Rural residence (v. urban) 0.11 (0.039)** 0.17 (0.033)*** −0.00 (0.064) 0.10 (0.039)**
 Immigrant (v. native born) −0.04 (0.076) 0.10 (0.075) −0.10 (0.150) −0.10 (0.070)
Economic resources
 Household wealth −0.07 (0.032)* −0.10 (0.026)*** −0.08 (0.052) −0.04 (0.032)
 Monthly income (colones in millions) −0.06 (0.036)+ 0.02 (0.042) −0.16 (0.104) −0.02 (0.072)
 Self-assessed economic situation 0.01 (0.023) 0.02 (0.02) 0.022 (0.035) 0.03 (0.023)
Social deprivation
 Currently unmarried (v. curr. married) 0.03 (0.038) 0.03 (0.034) 0.07 (0.065) 0.03 (0.043)
 Lives alone −0.02 (0.054) −0.02 (0.054) 0.15 (0.095) 0.06 (0.053)
 Low church attendance (< weekly) 0.03 (0.040) 0.04 (0.033) −0.04 (0.063) −0.05 (0.040)
Loss
 No. of children who have died (>= 1)c 0.05 (0.045) −0.00 (0.042) 0.03 (0.080) 0.05 (0.048)
Early childhood conditions
 Maternal age at death (years) −0.00 (0.001) −0.00 (0.002) 0.00 (0.002) 0.00 (0.001)
 Low maternal education (no edu.) −0.02 (0.046) −0.03 (0.039) −0.09 (0.075) −0.03 (0.043)
 Lived without biological father 0.01 (0.060) 0.01 (0.043) 0.03 (0.080) −0.02 (0.052)
 Poor health (>= 1 health problems) −0.09 (0.045)+ 0.05 (0.045) −0.01 (0.077) 0.01 (0.047)
 Economic deprivation index 0.02 (0.017) 0.05 (0.014)** −0.00 (0.027) 0.01 (0.016)
Open in a new tab

Note:

a

The regression coefficients are unstandardized, standard errors are inside the parentheses,

***

p-value < 0.001,

**

p-value < 0.01,

*

p-value < 0.05, and

+

p-value < 0.10.

Each column represents results from 17 separate regressions. The regressions for age and sex only control for sex and age, respectively.

b

All dependent variables have been logged (as per Table 1).

c

Regressions control for total number of children ever born.

fOne point toward index score for having mother with no formal education, mother who died before 50 years old, for growing up without a biological father, having reported one or more health problems early in life, and having reported three or more economic problems early in life.

Source: Authors’ calculations based on the 2004-2006 CRELES (Rosero-Bixby, 2007).

Table 4 presents estimated regression results for different variables and indices, with NAL as the dependent variable. A key finding from this table is the consistency and strength of the relationship between NAL and both age and female sex. Also importantly, most of the indices were not associated with NAL in the expected way. Most congruent with expectation is the positive correlation between the SES index and higher NAL values, significant at the 0.10 level. Although not significant at the 0.10 level, the social deprivation and loss indices do have the expected positive correlation with higher NAL levels.

Table 4. Estimated regression results from different models with neuroendocrine allostatic load (NAL) as the dependent variable, Costa Rica (ages 60 to 109, both sexes combined, years 2004-2006)a.

Dependent variable: NALb Model 1 Model 2 Model 3 Model 4 Model 5
Independent variables
Demographic
 Age 70-79 0.18 (0.065)** 0.18 (0.065)** 0.21 (0.065)** 0.21 (0.071)** 0.21 (0.071)**
 Age 80-89 0.41 (0.075)*** 0.40 (0.075)*** 0.45 (0.075)*** 0.39 (0.092)*** 0.40 (0.095)***
 Age 90-99 0.76 (0.159)*** 0.74 (0.161)*** 0.79 (0.158)*** 0.93 (0.262)*** 0.86 (0.272)**
 Age 100+ 0.81 (0.252)** 0.75 (0.242)** 0.81 (0.241)** 1.88 (0.478)*** 1.79 (0.504)***
 Female sex 0.41 (0.062)*** 0.41 (0.061)*** 0.42 (0.062)*** 0.43 (0.069)*** 0.41 (0.071)***
 Rural residence (v. urban) −0.02 (0.066) 0.01 (0.060) 0.04 (0.062) 0.06 (0.072) 0.05 (0.071)
 Immigrant (v. native born) −0.18 (0.113) −0.18 (0.110) −0.16 (0.109) −0.26 (0.128)* −0.23 (0.127)+
SES indexc 0.05 (0.028)+ -- -- -- 0.04 (0.032)
Social deprivation indexd -- 0.05 (0.038) -- -- 0.01 (0.045)
Loss
 No. of children who have died (>= 1)e -- -- 0.10 (0.072) -- 0.13 (0.082)
Early childhood conditions indexf -- -- -- −0.01 (0.034) −0.02 (0.034)
Constant 0.49 (0.064)*** 0.47 (0.067)*** 0.57 (0.067)*** 0.47 (0.072)*** 0.54 (0.087)***
N 1315 1329 1327 896 889
R2 0.089 0.089 0.092 0.084 0.093
Open in a new tab

Note:

a

The regression coefficients are unstandardized, standard errors are inside the parentheses,

***

p-value < 0.001,

**

p-value < 0.01,

*

p-value < 0.05, and

+

p-value < 0.10.

All regressions control for total number of children ever born.

b

NAL ranges from 0 to 4, with 4 representing highest risk. Results are presented with NAL cutpoints at the 25th and 75th percentiles.

c

One point toward index score for having low education, low wealth, low monthly income, and bad self-assessed economic situation.

d

One point toward index score for being currently unmarried, living alone, and low religious attendance.

e

Regressions with this variable in the model also control for total number of children ever born.

f

One point toward index score for having mother with no formal education, mother who died before 50 years old, for growing up without a biological father, having reported one or more health problems early in life, and having reported three or more economic problems early in life.

Source: Authors’ calculations based on the 2004-2006 CRELES (Rosero-Bixby, 2007).

We also examined only the currently married respondents to test whether having a spouse with low education or poor health was linked to worse biomarker profiles. Results (not shown) did not support the presence of these links. The last measure we analyzed was that of cumulative adversity -- in contrast to the stressors singly -- also in relation to individual biomarker and NAL values. In these regressions, included along with the overall stressor index variable were controls for age, sex, and children ever born. For the NAL construct, the coefficient was positive (i.e., more stressors were correlated with greater NAL values) and had a p-value of 0.182. Regarding the biomarkers analyzed individually, for cortisol, epinephrine, and norepinephrine riskier levels were associated with more stressors (p-values = 0.03, 0.365, and 0.221, respectively), but for DHEAS riskier levels were associated with fewer stressors (p-value = 0.002).

In addition to the results already presented and described, we ran a variety of additional analyses. These included using NAL cutpoints at the 10th and 90th percentiles instead of at the 25th and 75th, and also creating a NAL measure based on a summed z-score for respondents in which the score is the total number of standard deviations from the mean in the direction of high risk for each biomarker3. We also analyzed NAL as a binary variable (cutpoints NAL>=1 and NAL >= 2) and used logistic regression. These alternative methods of scoring the NAL construct produced largely the same results. The exceptions were that the relationship between NAL levels and low education, at least one child who has died, immigrant status, and the overall stressor index were considerably weaker. Separate analyses were also carried out by sex and using sex-specific biomarker cut points, and the results remain largely unchanged compared to the already presented combined-sex analyses. Lastly, since there is evidence to suggest that for cortisol, not only high, but low values as well, pose risk (Fries et al., 2005; Loucks, Juster, & Pruessner, 2008), we reran analyses examining both tails of cortisol’s distribution for the marker analyzed separately and as part of NAL constructs. The results of these analyses revealed that key associations such as between cortisol levels and both age and sex were attenuated, and in some cases especially so.

Discussion

This paper investigated stressors throughout the life course -- in early, middle, and later ages -- in relationship to riskier neuroendocrine biomarker profiles in a new, nationally-representative study of older Costa Rican men and women. The main finding is that contrary to our hypotheses, these stressors were not associated with individual biomarker levels and levels of NAL, a measure of neuroendocrine system dysregulation. Greater age and female sex, though, were linked to riskier biomarker and NAL values.

Findings and the wider literature

As just suggested, age and sex were the two lone characteristics in our study consistently and strongly correlated with our measures of neuroendocrine system dysregulation. These findings are not surprising. In the case of female sex, women tend to have lower levels of DHEAS than men (Goldman et al., 2004; Worthman, 2002), although evidence is mixed on whether they have higher resting levels than men for cortisol, epinephrine, and norepinephrine (Goldman et al., 2004; Hinojosa-Laborde et al., 1999; Van Cauter, Leproult, & Kupfer, 1996; Worthman, 2002). To the extent that women’s levels do differ, it seems due to some combination of greater stressor exposure, greater reactivity, and other predisposing pyschological and biological factors (Goldberg, 2006; Kajantie & Phillips, 2006; Piccinelli & Wilkinson, 2000).

In the case of age, since greater age can only bring about greater exposure to stressors, and since the allostatic framework theorizes that the costs to the body in dealing with challenge are cumulative, we would expect a positive relationship between age and NAL. In other words, that age and NAL are correlated with one another is only a necessary, but not sufficient condition for allostatic theory to hold. The challenge for the allostatic framework, then, is to demonstrate that when holding age constant, measures of a stressful life course are correlated with greater load.

In order to reconcile the findings here with the wider literature, we undertook an extensive literature search for articles related to linking stressors to levels of cortisol, DHEAS, epinephrine, norepinephrine, and AL. We focused on those articles which collected urinary samples to measure the catecholamines and cortisol, and blood samples to measure DHEAS. It is difficult to compare the findings in this paper with this wider literature because the latter seldom set out to test the hypothesis that stressors over the life course alter baseline levels of the neuroendocrine markers. Indeed, most of the studies that have attempted to link stressors to neuroendocrine marker levels have only examined one source (or few sources) of chronic stress. For instance, a study by Babisch and others (2001) tried to link traffic noise outside of respondents’ homes to catecholamine levels and another representative study examined whether women currently undergoing a divorce or separation had higher levels of the catecholamines and urinary free cortisol (Powell et al., 2002). In addition to the paucity of studies using multiple measures of stress over the life course, many did not include as many indicators of neuroendocrine system function as in the paper here and most made use of considerably smaller, non-population-based samples. Although it is difficult to summarize these varied papers, on the whole results appear mixed, with some supporting (Evans, 2003; Janicki-Deverts et al., 2007; Lemieux & Coe, 1995; Yehuda et al., 1995), some not supporting (Kubzansky et al., 1998; Powell et al., 2002) and others providing evidence for and against (Babisch, 2003; Luecken et al., 1997; Olff et al., 2006; Wheler et al., 2006) the connection between life stress and dysregulated neuroendocrine biomarker levels.

In addition to the studies already described, one stands out for its more thorough operationalization of chronic stress and ready comparability to the study here. This study is the Social Environment and Biomarkers of Aging Study (SEBAS) and it was carried out in Taiwan in 2000. Like the CRELES, the SEBAS is a large study (>1000 participants), is nationally representative, focuses on older persons (aged 54 and older), and has collected the catecholamines and cortisol through overnight urine samples and DHEAS through blood samples. In one study of the SEBAS by Gersten (2008b), he operationalizes the experience of stress over the lifecourse in part through the experience of such events as being widowed, living alone, and lack of group participation, as well as through respondents’ report of stress over their family’s work situation, health situation, marital situation, and other domains. Gersten fails to find a link between these stressors and the neuroendocrine markers analyzed in an index. This finding is supported by Dowd and Goldman’s analysis (2006) of the association between AL biomarkers and levels of education and income in the SEBAS. Glei and others (2007) and Goldman and others (2005) also investigate the connection between types of stress and AL levels in the SEBAS and find mixed results (although key parts of their analysis do not break up the findings by individual biomarkers or physiological systems and so it is difficult to directly compare their results to the ones in this paper).

Study limitations

One limitation of the survey analyzed here is the degree of proxy respondents (about 25%). This drawback is mitigated, however, by respondents’ high response rates, both in agreeing to take part in the interview portion (96%) and further willingness to give urine and blood samples (92 and 95%, respectively). For comparison, consider that for the SEBAS survey 92% gave interviews and 68% of these participants consented to the clinical examination (Goldman et al., 2003).

Since respondents in the CRELES were 60 and older, another limitation of the present study is respondents’ ability to remember events early in life. This issue may be most relevant for the questions inquiring about health problems in childhood and adolescence. It may be the case that respondents did experience the health problems asked about by surveyors, but did not know or did not remember the names of those problems. Imprecise recall seems less of an issue, however, for questions probing economic deprivation early in life, since it seems likely that respondents would be able to remember everyday events such as whether they grew up in a house with electricity, an indoor toilet, and whether they slept in a bed with others.

As mentioned in the introduction, a further weakness of the present study is that (except for a question about self-assessed economic situation) the study does not probe respondents’ subjective interpretations of their life history. Although we assume, for instance, that living alone is likely to be more stressful than not for most of the participants living alone, this may not indeed be the case. Nevertheless, the emotional response to certain human experiences, like the grieving involved in the loss of a child or spouse, seem close to being “universal,” and so it is still surprising that a number of variables that we investigated were not associated with our measure of physiological dysregulation. In the case of losing a spouse, not only is the loss itself psychologically difficult to deal with, but the loss could very well also result in future reduced instrumental and emotional support, thereby increasing stress levels for the widow or widower further.

Because one of the main aims of using biomarkers in a study such as the one here is to uncover the precise physiological mechanisms that underlie associations between social factors and health, it is reasonable to ask whether the independent variables we used in the analysis are indeed correlated with important health endpoints. In order to shed light on this question as well as provide some extra evidence that the independent variables we chose were indeed indicators of adversity, we ran additional analyses in which a variety of health outcomes were dependent variables and the independent variables were the ones used in this paper. We investigated the following health endpoints: self-rated health, physical frailty, ability to perform activities of daily living (ADL) and instrumental activities of daily living (IADL), cognitive disability, and depressive symptoms4. The results of these additional analyses are clear: a number of the independent variables used in the paper here are consistently and strongly related to the health measures just mentioned5. The variable most strongly and consistently related to the health endpoints was the cumulative adversity measure which was correlated with each of the five health outcomes in the expected way (i.e., a greater number of stressors was associated with worse health) and had p-values ranging from less than 0.001 to 0.0066.

To conclude, this is the first paper to use data from the CRELES, a nationally representative survey of older persons in Costa Rica, to attempt to link measures of emotional, social, and material resources (as well as negative life events and demands) to measures of physiological dysregulation. The negative findings in this paper raise doubts about a key assumption of the allostatic framework – that resting levels of the neuroendocrine markers become dysregulated through stress experienced over the life course.

Acknowledgments

Funding

The authors received the following financial support for the research and/or authorship of this article: The Wellcome Trust (award 072406/Z/03/Z) and the U.S. National Institute on Aging (award R01AG031716).

Footnotes

1

We confirmed that the missingness was not systematic by creating a dummy variable for whether respondents’ urine samples met the quality standards or not (for all those who gave samples). This dummy variable was then regressed on demographic characteristics (i.e., age, sex, educational level, rural residence, and immigrant status) and we examined results for both individual variables and tests of joint significance.

2

For example, the sample size of the CRELES for those with complete data on DHEAS, cortisol, epinephrine, and norepinephrine levels is 1,332, larger than that of comparable data for the SEBAS (n ~ 1,020) (Gersten, 2008) and the MacArthur studies (n ~ 870) (Seeman et al., 1997).

3

Unlike the cut-off approach, an index using the z-score method allows for unequal weighting of the biomarkers (e.g., a combined z-score of 3 could stem from being 2 SDs above the mean for cortisol, 1 SD above the mean for epinephrine, and the mean for the other two markers). The combined z-score is again the dependent variable in OLS regressions and can range from 0 to no pre-determined upper limit.

4

We chose these measures since they represent a wide variety of health outcomes (i.e., physical, cognitive, and psychological ones) and are based on self-report (e.g., self-rated health), answers to established scales (e.g., Yesavage’s symptoms of depression), and survey-administered tests (i.e., the physical frailty measure is based on performance on five separate physical tasks including grip strength, pulmonary peak flow, and chair stands. In operationalizing our health measures we follow the methods outlined in Rosero-Bixby and Dow (forthcoming) in which they also analyzed the CRELES data set.

5

For example, when using all available data and holding all other variables constant, low religious attendance was strongly related to physical frailty (p-value = 0.001), difficulty performing ADL and IADL (p-value = 0.004), poor self-rated health (p-value = 0.091), and depression (p-value = 0.001). In the same models, economic deprivation early in life was strongly associated with depression (p-value = 0.055) and poor self-rated health (p-values = 0.001). As might be expected, results were similar although attenuated when the sample was limited to those with complete data for cortisol, DHEAS, epinephrine, and norepinephrine (i.e., those respondents included in the NAL measure presented in the text).

6

Again, results were similar although attenuated for this relationship when the sample was limited to those with complete data for cortisol, DHEAS, epinephrine, norepinephrine.

Declaration of Conflicting Interests

The authors declared no conflicts of interest with respect to the authorship and/or publication of this article.

References

  1. Babisch W. Stress hormones in the research on cardiovascular effects of noise. Noise and Health. 2003;5(18):1–11. [PubMed] [Google Scholar]
  2. Babisch W, Fromme H, Beyer A, Ising H. Increased catecholamine levels in urine in subjects exposed to road traffic noise. The role of stress hormones in noise research. Environment International. 2001;26(7-8):475–81. doi: 10.1016/s0160-4120(01)00030-7. [DOI] [PubMed] [Google Scholar]
  3. Bähr J, Wehrhahn R. Life expectancy and infant mortality in Latin America. Social Science & Medicine. 1993;36(10):1373–1382. doi: 10.1016/0277-9536(93)90230-2. [DOI] [PubMed] [Google Scholar]
  4. Bolaños RL, Partanen T, Berrocal M, Alvárez B, Córdova L. Determinants of health in seasonal migrants: Coffee harvesters in Los Santos, Costa Rica. International Journal of Occupational and Environmental Health. 2008;14:129–137. doi: 10.1179/oeh.2008.14.2.129. [DOI] [PubMed] [Google Scholar]
  5. Brenes-Camacho G. Using geographical distance among family members from the electoral list as a measure of the intensity of social networks of the elderly in Costa Rica; Paper presented at the 2008 Annual Meeting of the Population Association of America in New Orleans; Louisiana. 2008. [Google Scholar]
  6. Budowski M, Rosero-Bixby L. Fatherless Costa Rica: child acknowledgment and support among lone mothers. Journal of Comparative Family Studies. 2003;34(2):229–254. [Google Scholar]
  7. Cacioppo JT, Hawkley LC. Social isolation and health, with an emphasis on underlying mechanisms. Perspectives in Biology and Medicine. 2003;46(3 Suppl.):S39–52. [PubMed] [Google Scholar]
  8. Cohen S, Kessler RC, Gordon LU. Strategies for measuring stress in studies of psychiatric and physical disorders. In: Cohen S, Kessler RC, Gordon LU, editors. Measuring stress: A guide for health and social scientists. Oxford University Press; New York: 1995. pp. 3–26. [Google Scholar]
  9. Crimmins E, Seeman T. Integrating biology into the study of health disparities. Aging, health, and public policy: Demographic and economic perspectives. In: Waite LJ, editor. Population and Development Review. Vol. 30. 2004. pp. 89–107. [Google Scholar]
  10. Crimmins EM, Seeman T. Integrating biology into demographic research on health and aging (with a focus on the MacArthur Study of Successful Aging) In: Finch CE, Vaupel JW, Kinsella K, editors. Cells and surveys: Should biological measures be included in social science research? National Academy Press; Washington, DC: 2001. pp. 9–41. [PubMed] [Google Scholar]
  11. Dow W, Schmeer KK. Health insurance and child mortality in Costa Rica. Social Science & Medicine. 2003;57(6):975–986. doi: 10.1016/s0277-9536(02)00464-1. [DOI] [PubMed] [Google Scholar]
  12. Dowd JB, Goldman N. Do biomarkers of stress mediate the relation between socioeconomic status and health? Journal of Epidemiology and Community Health. 2006;60(7):633–9. doi: 10.1136/jech.2005.040816. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Epel ES, Blackburn EH, Lin J, Dhabhar FS, Adler NE, Morrow JD, Cawthon RM. Accelerated telomere shortening in response to life stress. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(49):17312–5. doi: 10.1073/pnas.0407162101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Evans GW. A multimethodological analysis of cumulative risk and allostatic load among rural children. Developmental Psychology. 2003;39(5):924–33. doi: 10.1037/0012-1649.39.5.924. [DOI] [PubMed] [Google Scholar]
  15. Fries E, Hesse J, Hellhammer J, Hellhammer DH. A new view on hypocortisolism. Psychoneuroendocrinology. 2005;30(10):1010–6. doi: 10.1016/j.psyneuen.2005.04.006. [DOI] [PubMed] [Google Scholar]
  16. Gersten O. Neuroendocrine biomarkers, social relations, and the cumulative costs of stress in Taiwan. Social Science & Medicine. 2008a;66(3):507–519. doi: 10.1016/j.socscimed.2007.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gersten O. The path traveled and the path ahead for the allostatic framework: A rejoinder on the framework’s importance and the need for further work related to theory, data, and measurement. Social Science & Medicine. 2008b;66(3):531–535. [Google Scholar]
  18. Glei DA, Goldman N, Chuang Y-L, Weinstein M. Do chronic stressors lead to physiological dysregulation? Testing the theory of allostatic load. Psychosomatic Medicine. 2007;69(8):769–776. doi: 10.1097/PSY.0b013e318157cba6. [DOI] [PubMed] [Google Scholar]
  19. Goldberg D. The aetiology of depression. Psychological Medicine. 2006;36(10):1341–7. doi: 10.1017/S0033291706007665. [DOI] [PubMed] [Google Scholar]
  20. Goldman N, Glei DA, Seplaki C, Liu I-W, Weinstein M. Perceived stress and physiological dysregulation in older adults. Stress. 2005;8(2):95–105. doi: 10.1080/10253890500141905. [DOI] [PubMed] [Google Scholar]
  21. Goldman N, Turra CM, Glei DA, Lin Y-H, Weinstein M. Physiological dysregulation and changes in health in an older population. Experimental Gerontology. 2006;41(9):862–870. doi: 10.1016/j.exger.2006.06.050. [DOI] [PubMed] [Google Scholar]
  22. Goldman N, Weinstein M, Chang MC, Lin HS, Chuang YL, Lin YH, et al. 2000 Social Environment and Biomarkers of Aging Study in Taiwan (SEBAS): Main documentation for SEBAS public use data. 2003 Retrieved from http://www.icpsr.umich.edu.
  23. Goldman N, Weinstein M, Cornman J, Singer B, Seeman T, Chang M-C. Sex differentials in biological risk factors for chronic disease: estimates from population-based surveys. Journal of Women’s Health. 2004;13(4):393–403. doi: 10.1089/154099904323087088. [DOI] [PubMed] [Google Scholar]
  24. Hall C. Regional inequalities and well-being in Costa Rica. Geographical Review. 1984;74(1):48–62. [Google Scholar]
  25. Herring AA, Bonilla-Carrión RE, Borland RM, Hill KH. Differential mortality patterns between Nicaraguan immigrants and native-born residents of Costa Rica. Journal of Immigrant and Minority Health. doi: 10.1007/s10903-008-9121-y. In press. [DOI] [PubMed] [Google Scholar]
  26. Hinojosa-Laborde C, Chapa I, Lange D, Haywood JR. Gender differences in sympathetic nervous system regulation. Clinical and Experimental Pharmacology and Physiology. 1999;26(2):122–6. doi: 10.1046/j.1440-1681.1999.02995.x. [DOI] [PubMed] [Google Scholar]
  27. James GD, Brown DE. The biological stress response and lifestyle: catecholamines and blood pressure. Annual Review of Anthropology. 1997;26:313–335. [Google Scholar]
  28. Janicki-Deverts D, Cohen S, Adler NE, Schwartz JE, Matthews KA, Seeman TE. Socioeconomic status is related to urinary catecholamines in the Coronary Artery Risk Development in Young Adults (CARDIA) study. Psychosomatic Medicine. 2007;69(6):514–20. doi: 10.1097/PSY.0b013e3180f60645. [DOI] [PubMed] [Google Scholar]
  29. Kajantie E, Phillips DI. The effects of sex and hormonal status on the physiological response to acute psychosocial stress. Psychoneuroendocrinology. 2006;31(2):151–78. doi: 10.1016/j.psyneuen.2005.07.002. [DOI] [PubMed] [Google Scholar]
  30. Karlamangla AS, Singer BH, Greendale GA, Seeman TE. Increase in epinephrine excretion is associated with cognitive decline in elderly men: MacArthur studies of successful aging. Psychoneuroendocrinology. 2005;30(5):453–60. doi: 10.1016/j.psyneuen.2004.11.004. [DOI] [PubMed] [Google Scholar]
  31. Kubzansky LD, Berkman LF, Glass TA, Seeman TE. Is educational attainment associated with shared determinants of health in the elderly? Findings from the MacArthur Studies of Successful Aging. Psychosomatic Medicine. 1998;60(5):578–85. doi: 10.1097/00006842-199809000-00012. [DOI] [PubMed] [Google Scholar]
  32. Kubzansky LD, Kawachi I, Sparrow D. Socioeconomic status, hostility, and risk factor clustering in the Normative Aging Study: Any help from the concept of allostatic load? Annals of Behavioral Medicine. 1999;21(4):330–8. doi: 10.1007/BF02895966. [DOI] [PubMed] [Google Scholar]
  33. Lemieux AM, Coe CL. Abuse-related posttraumatic stress disorder: Evidence for chronic neuroendocrine activation in women. Psychosomatic Medicine. 1995;57(2):105–15. doi: 10.1097/00006842-199503000-00002. [DOI] [PubMed] [Google Scholar]
  34. Loucks EB, Juster RP, Pruessner JC. Neuroendocrine biomarkers, allostatic load, and the challenge of measurement. Social Science & Medicine. 2008;66(3):525–530. [Google Scholar]
  35. Low SM. The meaning of nervios: a sociocultural analysis of symptom presentation in San Jose, Costa Rica. Culture, Medicine and Psychiatry. 1981;5(1):25–47. doi: 10.1007/BF00049157. [DOI] [PubMed] [Google Scholar]
  36. Luecken LJ, Suarez EC, Kuhn CM, Barefoot JC, Blumenthal JA, Siegler IC, Williams RB. Stress in employed women: Impact of marital status and children at home on Neurohormone output and home strain. Psychosomatic Medicine. 1997;59(4):352–9. doi: 10.1097/00006842-199707000-00003. [DOI] [PubMed] [Google Scholar]
  37. McEwen BS. Protective and damaging effects of stress mediators. The New England Journal of Medicine. 1998;338(3):171–9. doi: 10.1056/NEJM199801153380307. [DOI] [PubMed] [Google Scholar]
  38. McEwen BS. Protective and damaging effects of the mediators of stress and adaptation: allostasis and allostatic load. In: Schulkin J, editor. Allostasis, homeostasis, and the costs of physiological adaptation. Cambridge University Press; Cambridge: 2004. [Google Scholar]
  39. Olff M, Güzelcan Y, de Vries GJ, Assies J, Gersons BP. HPA- and HPT-axis alterations in chronic posttraumatic stress disorder. Psychoneuroendocrinology. 2006;31(10):1220–1230. doi: 10.1016/j.psyneuen.2006.09.003. [DOI] [PubMed] [Google Scholar]
  40. Piccinelli M, Wilkinson G. Gender differences in depression. Critical review. British Journal of Psychiatry. 2000;177:486–92. doi: 10.1192/bjp.177.6.486. [DOI] [PubMed] [Google Scholar]
  41. Powell LH, Lovallo WR, Matthews KA, Meyer P, Midgley AR, Baum A, Stone AA, Underwood L, McCann JJ, Herro KJ, Ory MG. Physiologic markers of what chronic stress in premenopausal, middle-aged women. Psychosomatic Medicine. 2002;64(3):502–9. doi: 10.1097/00006842-200205000-00015. [DOI] [PubMed] [Google Scholar]
  42. Rosero-Bixby L. Biomarkers in a survey on aging in Costa Rica; Oral presentation at the International Union on the Scientific Study of Population (IUSSP) seminar on “Aging in developing countries: Building bridges for integrated research agendas.”; Santiago, Chile. April 23-24, 2007.2007. [Google Scholar]
  43. Rosero-Bixby L, Dow WH. Surprising SES gradients in mortality, health, and biomarkers in a Latin American population of adults. Journal of Gerontology: Social Sciences. doi: 10.1093/geronb/gbn004. Forthcoming. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Rosero-Bixby L, Dow WH, Lacle A. Insurance and other socioeconomic determinants of elderly longevity in a Costa Rican panel. Journal of Biosocial Science. 2005;37:705–720. doi: 10.1017/S0021932004006996. [DOI] [PubMed] [Google Scholar]
  45. Sandoval-García C. Threatening others: Nicaraguans and the formation of national identities in Costa Rica. Ohio University Press; Athens: 2004. [Google Scholar]
  46. Sapolsky RM. Why zebras don’t get ulcers. 3rd ed. Henry Holt & Co.; New York: 2004. [Google Scholar]
  47. Seeman TE, Glei D, Goldman N, Weinstein M, Singer B, Lin Y-H. Social relationships and allostatic load in Taiwanese elderly and near elderly. Social Science & Medicine. 2004;59(11):2245–57. doi: 10.1016/j.socscimed.2004.03.027. [DOI] [PubMed] [Google Scholar]
  48. Seeman TE, McEwen BS, Rowe JW, Singer BH. Allostatic load as a marker of cumulative biological risk: MacArthur studies of successful aging. Proceedings of the National Academy of Sciences of the United States of America. 2001;98(8):4770–5. doi: 10.1073/pnas.081072698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Seeman TE, Singer BH, Rowe JW, Horwitz RI, McEwen BS. Price of adaptation - allostatic load and its health consequences. MacArthur studies of successful aging. Archives of Internal Medicine. 1997;157(19):2259–68. [PubMed] [Google Scholar]
  50. Singer B, Ryff CD. Hierarchies of life histories and associated health risks. Annals of the New York Academy of Sciences. 1999;896:96–115. doi: 10.1111/j.1749-6632.1999.tb08108.x. [DOI] [PubMed] [Google Scholar]
  51. StataCorp. Stata statistical software: Release 9.0. StataCorp; College Station, TX: 2005. [Google Scholar]
  52. Steptoe A, Tsuda A, Tanaka Y, Wardle J. Depressive symptoms, socio-economic background, sense of control, and cultural factors in university students from 23 countries. International Journal of Behavioral Medicine. 2007;14(2):97–107. doi: 10.1007/BF03004175. [DOI] [PubMed] [Google Scholar]
  53. Thoits T. Stress, coping, and social support processes: Where are we? What next? Journal of Health and Social Behavior. 1995:53–79. Extra Issue. [PubMed] [Google Scholar]
  54. Timiras PS, Gersten O. The adrenals and pituitary -- stress, adaptation, and longevity. In: Timiras PS, editor. Physiological basis of aging and geriatrics. 4th ed. Informa Healthcare; New York: 2007. pp. 137–57. [Google Scholar]
  55. Turner RJ, Wheaton B. Checklist measurement of stressful life events. In: Cohen S, Kessler RC, Gordon LU, editors. Measuring stress: A guide for health and social scientists. Oxford University Press; New York: 1995. pp. 29–53. [Google Scholar]
  56. Turra CM, Goldman N, Seplaki CL, Glei DA, Lin Y-H, Weinstein M. Determinants of mortality at older ages: The role of biological markers of chronic disease. Population and Development Review. 2005;31(4):675–698. [Google Scholar]
  57. Van Cauter E, Leproult R, Kupfer DJ. Effects of gender and age on the levels and circadian rhythmicity of plasma cortisol. Journal of Clinical Endocrinology and Metabolism. 1996;81(7):2468–73. doi: 10.1210/jcem.81.7.8675562. [DOI] [PubMed] [Google Scholar]
  58. Wheler GH, Brandon D, Clemons A, Riley C, Kendall J, Loriaux DL, Kinzie JD. Cortisol production rate in posttraumatic stress disorder. The Journal of Clinical Endocrinology & Metabolism. 2006;91(9):3486–9. doi: 10.1210/jc.2006-0061. [DOI] [PubMed] [Google Scholar]
  59. Wolf AW, De Andraca I, Lozoff B. Maternal depression in three Latin American samples. Social Psychiatry and Psychiatric Epidemiology. 2002;37(4):169–176. doi: 10.1007/s001270200011. [DOI] [PubMed] [Google Scholar]
  60. Worthman CM. Endocrine pathways in differential well-being across the life course. In: Kuh D, Hardy R, editors. A life course approach to women’s health. Oxford University Press; Oxford: 2002. pp. 197–232. [Google Scholar]
  61. Yehuda R, Kahana B, Binder-Brynes K, Southwick SM, Mason JW, Giller EL. Low urinary cortisol excretion in Holocaust survivors with posttraumatic stress disorder. The American Journal of Psychiatry. 1995;152(7):982–6. doi: 10.1176/ajp.152.7.982. [DOI] [PubMed] [Google Scholar]

RESOURCES