Prior work has typically concerned diurnal cortisol rhythms over the course of the day, within-child child care variation across a more limited span , and rather blunt indices of child care exposure . In contrast, we considered children’s resting cortisol levels at a single point in time, within-child variation over the course of many months, and variation in the actual number of hours children spent in child care per week. Despite these differences , the commonalities between this prior work and the present findings are noteworthy. Taken collectively, the findings suggest that within-child increases in child care exposure—on both short- and long-term developmental time scales—tend to be associated with within-child increases in salivary cortisol, for children from low-risk contexts. In particular, in the present study we found that—holding all other time-invariant confounds constant—fairly long-term shifts in child care exposure across spans of several months may underlie changes in HPA axis function, at least at 24 months of age. This finding with respect to developmental timing is consistent with meta-analytic data suggesting that the link between child care and children’s cortisol increases over the course of the child care day tends to be quite small in infancy yet pronounced toward early childhood .
The cause of these developmental differences remains largely unclear. On the one hand, plastic growing bag it is reasonable to conjecture that the types of experiences that occur in child care, such as exposure to peers, noise, and the divided attention of caregivers, might be particularly salient stressors during infancy. Developmentally, infants lack the internal self-regulatory control that is presumably needed to down-regulate their stress responses . On the other hand, children’s experiences of the same child care context may also be quite different at different points in development . For instance, whereas an infant and early-childhood classroom may maintain the same number of peers, children tend to be more actively engaged with their peers in early childhood . Such normative social differences may begin to explain the emergence of the childcare effect in toddlerhood. This is an important question for future work. Critically, our findings indicated an inversed within-child relation between child care hours and cortisol at 24 month of age, for children from high-risk contexts. For instance, for toddlers with cumulative risk levels that were 1 SD above the mean, increases in weekly child care hours were associated with contemporaneous decreases in children’s cortisol levels. Although this conditional association was somewhat modest—corresponding to a standardized regression coefficient of approximately .14—it is approximately one quarter of the size of the effect of time of day, a known biological mechanism underlying within person cortisol variation. Furthermore, the estimate is robust as a within-person effect, which adjusts for all possible time-invariant confounds.
These findings align with our prior work with respect to testing between-child interaction effect as predictive of children’s resting cortisol levels at 48 months. In that work we found similarly that the direction of the association between child care hours and children’s cortisol levels differed for children from low- versus high-risk contexts . For children experiencing low levels of cumulative risk, attending a greater number of child care hours—on average, between 7 and 35 months—was predictive of comparatively higher cortisol levels, whereas for high-risk children a negative relation was evident. Indeed, the consistency between our prior and present findings is rather remarkable, given the different substantive interpretations of between- versus within-person effects. The former represents one’s average child care exposure compared to other children’s average exposures; the latter represents time-specific increases in child care hours at 24 months, compared to one’s own prior child care hours at 7 and 15 months—irrespective of one’s average level. It is important to note that our prior interactive findings at 48 months and the present interactive findings between 7 and 24 months almost certainly tap distinct effects. As above, the interpretations differ. In addition, 24-month and 48-month cortisol levels are virtually orthogonal . The present findings also align with prior studies that have considered interactive effects between environmental risk and child care exposure in the context of children’s behavioral outcomes. For example, using the same sample as the present study, in prior work we have found that, for children experiencing high levels of household chaos across early childhood, greater weekly hours in child care were predictive comparatively fewer behavior problems. In particular, consistent with the idea of a buffering effect, greater child care exposure ameliorated the detrimental relation between household chaos and children’s social problems.
Similar interactive effect have been noted by others , and work conducted with predominantly lower income samples has increasingly shown that the “beneficial” relations between child care and children’s social and academic outcomes may be stronger and more consistent than those from studies of children from more affluent families . Despite some commonalities emerging across this literature, the mechanisms underlying these interactions remain largely unclear. Our results provide some indication that center based care may play a role. We found that—for children from low-risk contexts—attending greater proportions of one’s time in child care in center-based care was predictive of higher cortisol levels than spending more time on other types of non-parental care. In contrast, no such relation was evident for children from high-risk contexts. To some degree, this is may be inconsistent with prior work. For instance, on the basis of data from their heterogeneous sample, Roisman and colleagues found that greater exposure to center-based care in early childhood was predictive of a lower awakening cortisol levels in adolescence. Of note, though, is that the cortisol awakening response is typically considered to be a process that is distinct from typical HPA axis functioning throughout the day . There is also, however, some reason to suspect that our findings align with some behavioral findings. For instance, there is a good indication that greater exposure to center-based care may be linked to heightened levels of aggression in early childhood —perhaps particularly so for children from more affluent families . Meta-analytic findings indicate a positive correlation between cortisol and aggression levels in early childhood . Indeed, considering the potential secondary effects of cortisol on behaviors and cognition is an important next step we intend to take with these data. However, there is also reason to suspect that the translation of children’s cortisol levels into behavioral effects will not be a simple one. For example, in the present study we also found that children who were exposed to greater numbers of peers, on average, wholesale grow bags between 7 and 24 months, tended to have lower cortisol levels. This is consistent with the extant data , yet it is somewhat inconsistent with the idea that the link between center-based care and heightened levels of aggression is explained by its effects on heightened HPA axis activity: Peer problems may well underlie the effects of center-based care on aggression . Suffice it to say, there is much left to clarify. Interestingly, prior work has been suggested that the negative relation between greater peer play and lower daily cortisol levels may reflect the fact children who are less physically aroused may “select into” greater amounts of peer play . Of course, this may well be the case, yet it is noteworthy that our finding with regard to the number of children in child care—presumably reflecting peer exposure that is considerably less child driven—shows similar negative relation. Although it remains a conjecture to be tested directly, ongoing practice negotiating the social world of one’s peers may, in fact, play a “steeling” or “inoculating” role in the way that children respond physiologically to subsequent peer interactions. Evidence from nonhuman primates suggests that such steeling effects may be common and play an important role in resiliency .As such, our present findings suggest that child care type and peer exposure—at least when considered cumulatively between ages 7 and 24 months—are linked with HPA axis functioning across this period.
However, it is important to note that they do not appear to explain the effects of childcare quantity. Child care quantity, type, and peer exposure were additively and independently predictive of children’s cortisol levels, and no high-order interactions were evident. Like our prior work with children’s 48-month cortisol levels, caregiver responsivity was not related with children’s cortisol levels, between or within persons. Thus, as noted by others in prior work the mechanism underlying the child care quantity effect remains largely unclear. One possibility is that, for children facing the confluence of environmental challenges of poverty, child care may be comparatively less physiologically stressing than their experiences outside of child care. Indeed, the limited available evidence suggests that— unlike their more affluent peers—children from higher risk contexts tend to show a normative diurnal cortisol decline in their cortisol levels across the child care day . If this is the case, then the “beneficial” relation between greater child care exposure and cortisol observed for children from high-risk families may partially reflect a dose–response relation, whereby greater child care exposure leads to less exposure to the pronounced environmental stressors outside of child care. This, of course, remains a question to be addressed directly. Indeed, there are several potentially important mechanisms that were not addressed in the present study—such as the quality of children’s peer relations , teacher–child relationship quality , and degree of classroom chaos —that remain critical areas inquiry, with regard to clarifying the relation between child care exposure and HPA axis functioning.Several of our findings are robust to within-child analyses that accounted for all time invariant confound or between-child analyses that adjusted for several observed control covariates. In fact, with regard to within-child effects, the conservative nature of these estimates may underlie the fact that there were several null within-person relations . On the one hand, these specifications may strengthen the internal validity of our inferences. On the other hand, to the extent that the causal relation between these variables is contingent on cumulative exposure they would be missed by our within-person specification. They allow us to pick up cumulative, traitlike variability in these child care measures—variability that is likely quite meaningful. However, our inferences are only causally unbiased to the extent to which we have accounted for all sources of potential endogeneity . This assumption is typically dubious. As such, we make no causal claims. Nontrivially, though, these potential endogeneity problems would have to explain both selection into cumulative risk and selection into child care, as well as why the direction of the child care effects vary as a function of cumulative risk.As is common with large, comprehensive epidemiological deigns, trade-offs must often be made between breadth and depth. For instance, we were unable to collect multiple saliva samples over the course of a day or across several consecutive days. This would have strengthened the reliability of our cortisol measures and made our findings more directly informative to the extant literature concerning children’s diurnal cortisol patterns . Our saliva samples were collected modally around 10:00 a.m., when the child was at rest, after he or she had had time acclimate to the RAs in the home. There was little evidence that children were particularly stressed by the RAs’ visits. The vast majority of mothers rated their children’s behavior during the visit as being typical. Of the minority who indicated that their children were behaving differently, it was typically due to external reasons . For the few who ascribed differences as being due to the RAs’ visit, they were evenly split between those showing emotions that were comparatively more negative and comparatively more positive than usual. This is consistent with the extant laboratory-based literature that suggests that exposure to novel adults typically fails to evoke an acute stress response—even when it is intended to do so . That being said, we cannot rule out the possibility that our cortisol measures may partially tap reactivity to the RAs’ presence. Finally, by using terms like beneficial to describe, for instance, the negative relation between child care hours and cortisol, we imply that lower cortisol levels may be substantively “better” than higher cortisol levels. Other interpretations are possible. For instance, high levels of cumulative risk could make these children especially vulnerable to child care effects that cause the HPA axis to be underactive . Underactive HPA axis functioning is also a known risk factor for less optimal cognitive functioning and behavior problems . We cannot rule this interpretation out; however, we find it less plausible.