A longitudinal study of basal cortisol in infants: Intra-individual variability, circadian rhythm and developmental trends

https://doi.org/10.1016/S0163-6383(02)00141-8Get rights and content

Abstract

Mothers with normally developing babies were visited in their homes during 13 consecutive weeks, when the babies were around 5–8 months of age. Basal salival cortisol measures were taken for both the baby and the mother on arrival.

The infants’ basal cortisol decreased linearly with age, was negatively related to sleep, and did not show adult-like circadian declines from morning to mid-afternoon. Furthermore, while the infants showed relative stability across individuals, they displayed great intra-individual variability across assessments. Contrarily, the mothers displayed important inter-individual variability, together with a relative stability across assessments. The infants’ important intra-individual variability was not affected by gender, nor time of visit, nor was it related to the mothers’ basal cortisol. Daily measures of basal cortisol taken in a subgroup of infants indicated the day-to-day intra-individual variability to be of the same magnitude as the week-to-week variability.

The question of how the intra-individual variability in basal cortisol affects assessments of cortisol in infancy is addressed. The aggregation of data with the goal of increasing the reliability of the assessments is shown to be inadequate for infant basal cortisol.

Introduction

Cortisol is a steroid hormone belonging to the hypothalamic-pituitary-adrenocortical axis. It is secreted in a pulsatile fashion by the adrenal cortex and its levels show a strong circadian rhythm, being lowest around midnight and highest in the early morning hours (Kirschbaum & Hellhammer, 1989).

In recent years, an upsurge in studies of cortisol response has occurred mainly due to two reasons. First, cortisol is a hormone with many different and important functions, which are mainly aimed at regulating circadian-driven activities (de Kloet, Rosenfeld, van Eekelen, Sutanto, & Levine, 1988). Cortisol intervenes in energy release (e.g., liberation of glucose, inhibition of insulin), immune activity (e.g., downregulating inflammatory responses and the cytokine cascade), mental activity (e.g., increased alertness, memorization, learning), growth system (e.g., inhibition of growth hormone), and reproductive function (e.g., inhibition of gonadal steroids) (Flinn & England, 1997). As such, it is an essential hormone for the energy balance in our bodies, producing a broad spectrum of physiological effects as virtually all body cells are affected by cortisol (Kirschbaum & Hellhammer, 1989). Cortisol also plays an important role in stress responses, as it is secreted when the organism faces a difficult or problematic situation that is perceived as stressful.

A second reason for the success in being the object of scientific endeavor is due to practical laboratory developments of the last decades (Kirschbaum & Hellhammer, 1989, Kirschbaum & Hellhammer, 1994). Previously cortisol levels could only be assessed in serum or urine. Currently, cortisol can be easily and reliably assessed in saliva using relatively cheap and ready-made kits (Schwartz, Granger, Susman, Gunnar, & Laird, 1998).

An area that has especially benefited from the technological advances is the study of stress in infants and children. In addition to behavioral and psychological assessments, physiological measures are often added to infant research designs to provide a more objective measure (Spangler & Scheubeck, 1993). Thus, in addition to its physiological importance, the technical accessibility and the popularity of inter-disciplinary studies, have made cortisol a much-measured hormone in developmental studies (e.g., Gunnar, Brodersen, Krueger, & Rigatuso, 1996; Larson, White, Cochran, Donzella, & Gunnar, 1998; Lewis & Ramsay, 1995; Spangler & Grossmann, 1993). Examples of such infant studies are those studying behavioral and hormonal reactions to potentially stressful situations such as maternal separations, inoculations, new situations and physical examinations (Gunnar et al., 1996; Gunnar & Donzella, 2002; Gunnar, Larson, Hertsgaard, Harris, & Brodersen, 1992; Hertsgaard, Gunnar, Larson, Brodersen, & Lehman, 1992; Larson et al., 1998; Lewis & Ramsay, 1995; Spangler & Scheubeck, 1993). Cortisol measures have also been taken during attachment assessments, in which infants are subjected to a potentially stressful event, namely the temporal separation from their caretaker (Goldsmith & Harman, 1994; Gunnar, Mangelsdorf, Larson, & Hertsgaard, 1989; Hertsgaard, Gunnar, Erickson, & Nachmias, 1995). One or more measures of cortisol are typically taken after the potentially stressful situation, and then compared to a measure of “basal” or resting cortisol taken before the onset of the situation. This first sample is taken on arrival at the site where the stressful situation will take place. Longitudinal studies of the developmental course of the reactions typically collect two to four consecutive measurements. Given the importance of the basal cortisol level as a point of reference, it is somewhat surprising to see that the picture of the development of basal cortisol in infants is still far from complete.

Although we speak about “basal” cortisol, suggesting that the cortisol hormone level is stable, the truth is that basal cortisol varies to a considerable extent. This variation depends on various causes and differs over various time intervals. On the level of time intervals, basal cortisol varies over the relatively short interval of a day in the form of circadian rhythm. It also varies over the longer time interval of age. In addition, basal cortisol depends on preceding conditions such as sleep or feeding, particularly in infants, but also on daily experiences or physical condition. The issue of variation over age is closely related to the fact that basal cortisol is subject to development. However, development is broader than just age changes, because it also affects circadian rhythm and the nature of the conditional factors.

As mentioned earlier, an important characteristic of the cortisol hormone is that it displays a circadian rhythm. This daily variation in basal levels is well documented in adults (Van Cauter & Turek, 1995). Infants, in contrast, are born without a circadian rhythm in cortisol but they acquire it during their first months of life. Studies differ in the exact age of appearance of the circadian rhythm: from as early as 2 months till the age of 9 months (Antonini, Jorge, & Moreira, 2000; Kiess et al., 1995; Lewis & Ramsay, 1995; Mantagos, Moustogiannis, & Vagenakis, 1998; Onishi et al., 1983; Price, Close, & Fielding, 1983; Santiago, Jorge, & Moreira, 1996; Spangler, 1991). Also, even when the circadian rhythm has been found to be present, the nature of the rhythm has been found to be different than that of adults. Gunnar and Donzella (2002), for example, conclude that significantly lower mid-afternoon than mid-morning levels (which are characteristic of adult curves) cannot be obtained reliably until children are around 4 years of age.

Another interesting question is whether we can find changes over age in basal cortisol levels and in reactions to stressors, especially in infants. As Gunnar et al. (1996, p. 878) state: “  data on the development of cortisol responses are needed to inform both empirical and theoretical work.” This is especially true given the contradictory nature of some earlier findings.

Earlier research has found a dampening in cortisol response to stressors from around 3–4 months of age till at least 15 months of age (Gunnar et al., 1996, Larson et al., 1998; Lewis & Ramsay, 1995). At the same time, no changes in basal cortisol were found between 7 and 15 weeks of age (Larson et al., 1998) and between 2 and 6 months of age (Gunnar et al., 1996), while a decrease in basal cortisol was found between two assessments at 6 and 15 months of age (Gunnar et al., 1996). Contrarily, Lewis and Ramsay (1995) found a decline in cortisol levels between 2 and 6 months of age, and Price et al. (1983) report a decrease in daily mean values between 1 and 20 weeks of age.

One of the problems with the basal cortisol samples taken in all the studies except Price et al.’s is that they are actually pre-test values, measured in a clinic or laboratory. As such, they may in part be influenced by other stressors or variables (Hertsgaard et al., 1992; Larson, Gunnar, & Hertsgaard, 1991; Lewis & Ramsay, 1995). Also, the studies are often cross-sectional or consist of few assessment ages per infants.

In general, basal cortisol levels are affected by various situational factors, such as daily hassles, physical condition, social factors, etc. (Ehlert, Patalla, Kirschbaum, Piedmont, & Hellhammer, 1990; Flinn & England, 1995, Hellhammer & Wade, 1993; Kirschbaum, Bartussek, & Strasburger, 1992). Thus, when assessing an individual’s basal cortisol, the value obtained will be related to various external and internal factors. Accordingly, it will vary daily with the individual’s situation. In adults, considerable day-to-day intra-individual variability in cortisol concentrations has been found (Kirschbaum & Hellhammer, 1989, Kirschbaum et al., 1990). Adult basal cortisol tends to be more stable and have more predictive validity with regard to socioeconomic and personality variables, and age, when it is taken in the morning hours (Brandtstädter, Baltes–Götz, Kirschbaum, & Hellhammer, 1991). In infants, basal cortisol is affected by two particular situational factors, namely sleep and feeding.

Spangler (1991) found no relation between mean cortisol levels and the mean frequency of sleep and the mean duration of sleep per day and per episode. He did however find that cortisol values were higher when a high amount of sleep had occurred in the hours preceding the assessment. Lewis and Thomas (1990) found similar associations: a negative correlation between cortisol level and the time awake for 6-month-old infants (but not for 2- and 4-month-olds). On the other hand, Larson et al. (1991) found morning naps to be associated with decreases in cortisol in 9-month-old infants with returns to pre-nap levels 45 min after waking up.

High cortisol levels have been found to be associated with a higher feeding latency (Gunnar, Malone, Vance, & Fisch, 1985). This is supported by Spangler (1991), who found first that a higher frequency of feeding episodes was associated with somewhat lower mean cortisol values, and second that cortisol levels decrease immediately after a feed and increase later on. Feeding behavior thus seems to inhibit adrenocortical activity (Levine, Coe, & Wiener, 1989). However, there are studies that conclude that there appear to be no differences in pre- and post-feeding salivary cortisol levels (Lobo, 1990), while others state that the time elapsed since the last solid feeding (but NOT since the last liquid feeding) seems to correlate with basal levels of cortisol (Hertsgaard et al., 1992). This last finding about solid feeding is probably due to the post-prandial surge in cortisol which occurs approximately 45 min after the noonday meal (Riad-Fahmy, Read, & Hughes, 1983).

The sensitivity to situational factors will lead to considerable intra-individual variability in cortisol levels, depending on which factors affect which individuals and when. Since various aspects of basal cortisol production are subject to development, we may ask ourselves whether intra-individual variation itself is a developmental phenomenon. Newborns show high day-to-day and within-day stability in cortisol levels, which will at least in part be due to their lack of a circadian rhythm at birth (Spangler, 1991). This makes the assessment of cortisol in newborns reliable. The question remains whether this situation is the same in older infants. In other words, how does the intra-individual variability in basal cortisol affect assessments of cortisol in infancy? This question is important because answering it will tell us just how reliable measures of basal cortisol in infants are.

From the preceding overview we may conclude that there are many factors affecting variations in basal cortisol in both adults and infants. As far as infants are concerned, our knowledge of major sources of variation—circadian rhythm, age, situational factors—is still far from unequivocal. An important issue, both from a developmental and a methodological viewpoint, is that of variability. Earlier research has indicated the possibility that high intra-individual variation is a natural feature of normal development in young infants. In a longitudinal study of emotion-related behaviors in infants we found major intra-individual variability during the first 10 months of life, together with reduced variability thereafter (de Weerth, van Geert, & Hoijtink, 1999; de Weerth & van Geert, 2000). Thus, as the infant develops, the intra-individual variability is reduced, apparently giving way to a narrower individual equilibrium range. Does this pattern also apply to basal cortisol, i.e., is basal cortisol stable enough to count as a characteristic property of the individual infant? Or is the intra-individual variation in cortisol comparable to variation between individual infants? If the intra-individual variability of basal cortisol is different for infants and adults, how does variability change in the course of early development?

In this study we have attempted to answer these questions by assessing the variability of basal cortisol in an intensive longitudinal fashion in both mothers and 5- to 8-month-old infants. Because mothers and their infants are genetically related and are exposed to more or less the same family environmental stressors (Spangler, 1991), they constitute nicely matched pairs in which to study adult–infant differences in variability of basal cortisol. In this way, the intra-individual variability in basal cortisol can be compared to the inter-individual variability in both the mothers and the infants. Also, an attempt was made to determine whether infant basal cortisol varies on a daily or weekly basis, and whether it is influenced by infant gender and/or by maternal cortisol values (which represent genetic and environmental influences). These factors have been related to infant-cortisol levels in the past (Brandtstadter et al., 1991, Gunnar et al., 1996; Kirschbaum, Wust, Faig, & Hellhammer, 1992; Kirschbaum, Wust, & Hellhammer, 1992). We have also tried to determine whether the daily circadian variations of cortisol displayed by young infants are similar to those shown by adults. Finally, we examined the possibility of aggregating data in order to increase reliability.

Summarizing, the major aim of this study was to examine measures of intra-individual variability in infant basal cortisol. Our additional goals are to explore evidence relating to circadian rhythm and to investigate the occurrence of developmental trends in basal cortisol levels.

Section snippets

Subjects

The subjects were 20 infants (8 males and 12 females) born after singleton, uncomplicated pregnancies and vaginal deliveries. One infant (pair 1) was born 23 days before the due date, but because she was healthy, did not need to be put in the incubator and had Apgar scores of 9−10 at birth, she was included in the study. The remaining infants were born after pregnancies of 37−42 weeks and also had normal Apgar scores at birth. The mean birth weight was 3,550 g and the mean birth length was 50.8 

Statistical analysis

A preliminary, descriptive data analysis was carried out with the aid of non-parametric exact methods, based on permutation and bootstrap techniques. These computationally intensive methods provide exact measures of p-values independent of assumptions regarding the distribution of the data and allow one to estimate the confidence intervals of test statistics such as means, standard deviations and correlation coefficients (Efron & Tibshirani, 1993, Good, 1999).

Further analysis of the data was

Intra-individual variability

The main question we attempted to answer in this longitudinal study concerned the nature of intra-individual variability in basal cortisol levels. According to the literature, considerable day-to-day intra-individual variability in cortisol is to be expected in adults (Kirschbaum & Hellhammer, 1989, Kirschbaum et al., 1990).

The adult females who were assessed in our study showed variability in their weekly cortisol levels. The reliability for the assessment of maternal cortisol levels was 61%,

Acknowledgements

This research was funded by the Netherlands Organization for Scientific Research (NWO), grant number 575-63-078. We wish to thank T. Snijders for his statistical advice, D. Kroeze and R. van Bruggen for collecting the data, and also the families that participated in the study.

References (53)

  • de Kloet, E. R., Rosenfeld, P., van Eekelen, J. A. M., Sutanto, W., & Levine, S. (1988). Stress, glucocorticoids and...
  • de Weerth, C. (1998). Emotion-related behaviors in infants. A longitudinal study of patterns and variability. Ph.D....
  • C de Weerth et al.

    Intra-individual variability in infant behavior

    Developmental Psychology

    (1999)
  • de Weerth, C., & van Geert, P. (2000). The dynamics of emotion-related behaviors in infancy. In M. D. Lewis & I. Granic...
  • de Weerth, C., van Zijl, R., & Buitelaar, J. K. (2002). Longitudinal study of the appearance of the salivary cortisol...
  • Efron, B., & Tibshirani, R. (1993). An introduction to the bootstrap. New York: Chapman and...
  • M.V Flinn et al.

    Childhood stress and family environment

    Current Anthropology

    (1995)
  • M.V Flinn et al.

    Social economics of childhood glucocorticoid stress response and health

    American Journal of Physical Anthropology

    (1997)
  • H.H Goldsmith et al.

    Temperament and attachment individuals and relationships

    Current Directions in Psychological Science

    (1994)
  • Goldstein, H., Rasbash, J., Plewis, I., Draper, D., Browne, W., Yang, M., Woodhouse, G., & Healy, M. (1998). A user’s...
  • Good, P. I. (1999). Resampling methods: A practical guide to data analysis. Boston:...
  • M.R Gunnar et al.

    Dampening of adrenocortical responses during infancy: Normative changes and individual differences

    Child Development

    (1996)
  • M.R Gunnar et al.

    The stressfulness of separation among 9-month-old infants: Effects of social context variables and infant temperament

    Child Development

    (1992)
  • M.R Gunnar et al.

    Coping with aversive stimulation in the neonatal period: Quiet sleep and plasma cortisol levels during recovery from circumcision

    Child Development

    (1985)
  • M.R Gunnar et al.

    Attachment, temperament, and adrenocortical activity in infancy: A study of psychoendocrine regulation

    Developmental Psychology

    (1989)
  • M.R Gunnar et al.

    Event-related potentials in year-old infants: Relations with emotionality and cortisol

    Child Development

    (1994)
  • Cited by (0)

    View full text