We explored in a socioeconomically at-risk cohort of urban-dwelling older adults enrolled in a large-scale trial of a social health promotion program the baseline associations between measures of diurnal variation in cortisol and SES. Whether classifying participants according to perceived location in the lower vs. upper half of the SES ladder or by combined current household income of above and below $25,000, we observed blunted decline in diurnal cortisol among participants in the lower vs. higher SES strata. These associations were unaltered when adjusting for age, sex, and BMI. We did not observe SES-related differences in awakening cortisol, CAR, or AUC. Our results confirm studies reporting a blunted decline in diurnal cortisol among lower vs. higher SES groups22–24
, extending this work to an older, community-dwelling sample of socioecomically at-risk adults. They suggest that one major biologic mechanism through which environmental disadvantage accumulates is through dysregulation, e.g. chronic elevation, in daily cortisol levels as demonstrated here by blunted diurnal decline.
Blunted diurnal cortisol appears to be a sensitive measure of decreased physiologic resilience under the concepts of frailty and disability, and, allostatic load (AL) and risk for chronic disease. Frailty has been conceptualized as a disease-independent state of impaired regulation across multiple systems and is associated with increased vulnerability to physical disability.25–27
In earlier work among older women, we provided the first epidemiologic evidence that higher levels and blunted diurnal cortisol were associated with increasing frailty burden, whereas awakening was not.28
AL has been described by Seeman et al.
as a biological pathway through which environmental risk factors over the life course cumulatively impose a significant physiological burden on multiple, interrelated body systems.29–31
Biological evidence for the deleterious effects of over-stimulation of the HPA-axis can be observed using numerous markers of brain health, including atrophy, deactivation, and cell death in the hippocampus 32
, a region important to memory and risk for Alzheimer’s Disease. Behaviorally, healthy older adults with increasing and elevated cortisol levels over the preceding four years exhibited poorer memory and attention relative to those with lower levels.33
Elevated and increasing levels of cortisol were also associated with reduced hippocampal volumes.34
These effects may be compounded, or exacerbated, in those experiencing chronic environmental disadvantage through low SES.
The socioeconomic composition of older adults participating in the BECT offers a critical opportunity to examine the intersection of aging and life time cumulative risk through low SES, given that these individuals are at elevated risk for disease and disability. Education was variable with 37% having a high school education or less. The median total income before taxes in our sample was $30,000, and often represents combined income in many participants who report living with others. The federal poverty guidelines commonly used by the Department of Health and Human Services (HHS) (2011) for a family of four is $22,350. In addition, the majority of participants are African-American, and represent a segment of the population experiencing increased financial strain that likely affects health.35
In this group, we nonetheless found numerous similarities across socioeconomic strata in diurnal cortisol levels for awakening, CAR, and AUC. Specifically, we did not observe a difference in awakening cortisol level according to perceived SES or combined household income. Several studies have reported higher awakening cortisol levels among high SES participants8, 9, 36
although few studies have reported the opposite association.37, 38
Other studies, like ours, reported no difference in awakening cortisol by SES group.14, 15, 37, 39, 40
These equivocal findings may reflect variability in socioeconomic strata studied, and variable sleep-wake patterns, as observed here with wake times ranging from 4:00 a.m. to 10:00 a.m. Similarly, for the CAR, results are inconclusive with some finding an association with SES.14, 37, 41
We did not observe a difference in CAR by perceived SES and by total current household income, confirming reports from several other studies.14, 36, 37, 41
Whereas a lower CAR has been associated with chronic health problems, posttraumatic stress disorder, chronic fatigue syndrome, and sleep disorders, day-to-day fluctuations in CAR have been observed.42
Finally, AUC is a commonly used measure of total cortisol secreted with studies finding that lower SES was significantly related to a higher AUC.15, 17, 43
However, our study showed no association between AUC and SES. This may be due, in part, to limited sensitivity of the AUC when two participants have the same AUC value but different patterns of diurnal secretion that may suggest chronically elevated levels.
An important factor in the assessment of diurnal cortisol that may partially explain equivocal findings is the variability in and accuracy with which one takes and reports collection at the indicated times. Differential adherence to timing contributes to artificial blunting of the diurnal cortisol pattern.7, 41
However, few investigators report compliance rates in the collection of samples at designated times and this has a significant impact on the reported results.7, 13, 14, 41, 43, 44
In modeling the data, we observed great variability in sample collection times and recognized the potential for measurement error by assuming fixed collection times. Therefore, we applied one fitted regression slope and incorporated time administered for samples 2–4, thus reducing the potential for measurement error and increasing sensitivity in detecting associations between SES and diurnal cortisol (see methods). The methodologic approach used here to model diurnal decline in cortisol represents a novel metric intended to mitigate the effect of this unmeasured variability and a represents a particular strength of the study.
This study identified biologic vulnerability among individuals with lower SES who are participating in a two-year randomized trial studying the effects of activity and environmental enrichment through volunteer service to elementary school children. Having collected annual follow-up cortisol samples over the two years of the BECT, we will now be able to examine whether this biologic link to chronic environmental deprivation and disadvantage over the life course can be partially reversed when these individuals volunteer in an enriched environment. Specifically, we will extend this work to determine how such vulnerability over the life course may be partially reversed through these same plastic regulatory pathways. Animal models show that enriched environments elicit neurogenesis or synaptogenesis (i.e., growth of new neurons or formation of new synapses between neurons), and reduced neuronal death, especially in hippocampal structures important to memory 45
and dementia risk. There is ample evidence of these neuroplastic changes in older animals.46–50
However, this work has only recently begun to be translated to human studies of chronic stress 51
, and remains to be explored in older adults who may have experienced decades of accumulated physiologic burden in multiple, interrelated systems.
There are a number of study limitations that warrant consideration. First, while the study sample was relatively small (N
<100), but was nonetheless comparable to other studies examining diurnal cortisol in older adults and was sufficiently powered to observe SES-related differences in diurnal decline using two different measures. Second, our collection of cortisol samples was restricted to one day unlike other studies that obtained samples over three or more days.14, 15, 52, 53
To address this limitation, we are currently administering a validation study to assess cortisol concentrations over two days. Third, we did not adjust for the day of the week during which samples were obtained, as daily patterns may be less variable in retired adults. Nonetheless, this limitation suggests that our results may represent a conservative estimate of the association between cortisol and SES. Finally, regarding genearlizability, these findings were from a study of participants willing to enroll in a two-year trial of high-intensity volunteer service. These individuals may represent the health-conscious members of their community, which again would lead to a conservative estimate of the association between SES and biologic vulnerability as measured by diurnal cortisol.
We used perceived position on the SES ladder as an SES measure to complement current total income; because most studies exploring this association have not used the SES ladder, we were unable to compare our results to multiple studies. However it should be noted that utilization of the SES ladder provides more information about the social status of an individual across multiple SES indicators compared to education and or current income especially in an older population.16