Diurnal salivary cortisol levels in postpartum mothers as a function of infant feeding choice and parity

Summary

Daily stress and sleep deprivation can influence the diurnal pattern of cortisol, which normally consists of high morning levels and a gradual decline throughout the day. While most individuals have consistent declining cortisol concentrations over days, others display either flat or inconsistent profiles. Postpartum mothers experience considerable home demands and sleep deprivation, yet, breastfeeding mothers perceive lower stress and reduced negative mood states compared to bottlefeeders. On the other hand, multiparity (having more than one child) is associated with reduced steepness in diurnal cortisol decline. Interestingly, no study to date has investigated the diurnal cortisol pattern and its stability across days in postpartum women as a function of their choice of infant feeding and parity. In this study, we measured salivary cortisol at four different time points during the day, on two non-consecutive days in first-time (primiparous) and second-time (multiparous) mothers at 5–20 weeks postpartum who were exclusively breastfeeding or bottlefeeding, and in non-postpartum mothers of young children (1–6 years). Among multiparous mothers, we found that cortisol levels in those who were bottlefeeding were higher than in breastfeeding mothers at both awakening and 1600 h. This effect remained significant after controlling for individual differences related to infant feeding choice, such as estradiol levels, education and income. No effect of infant feeding choice on cortisol concentrations was observed in primiparous mothers. While a consistent decline across days was common, some mothers presented a flat or inconsistent profile, a profile that was not associated with infant feeding choice or parity. Importantly, mothers with consistent declining profiles had the highest household income. Our findings suggest that although breastfeeding might promote a tighter regulation of diurnal basal cortisol secretion, in particular for multiparous mothers who are likely to be exposed to greater home demands and maternal responsibilities, some aspects of socioeconomic status such as income can also play a significant role in the stability of diurnal cortisol secretion across days.

Introduction

In humans, the postpartum period is generally associated with significant lifestyle changes related to infant care, which can induce stress and sleep disruption. In turn, these changes can alter the circadian secretion of cortisol, the end-product of the hypothalamic–pituitary–adrenal (HPA) axis. Indeed, cortisol secretion, which normally peaks during the early morning hours and gradually decreases throughout the day (Curtis and Fogel, 1971, Cox et al., 1987, Weitzman et al., 1971), can significantly increase in response to acute stressors occurring during the day (Smyth et al., 1998, van Eck et al., 1996). In addition, both acute sleep deprivation and chronic primary insomnia can lead to greater evening cortisol levels (Ilias et al., 2002; Leproult et al., 1997; Vgontzas et al., 2001).

Infant feeding choice, as well as related secondary factors such as estrogen levels and socioeconomic status (SES), might also affect basal cortisol levels in postpartum mothers. For instance, an acute suckling episode is associated with reduced cortisol secretion in the following 15 min (Amico et al., 1994), yet the cumulative effect of repeated nursing on diurnal cortisol secretion is not known. Circulating estrogen levels, which are greater in bottlefeeding mothers, are known to enhance cortisol secretion (Magiakou et al., 1997). Finally, high maternal education in women is a strong predictor of choosing breastfeeding over bottlefeeding as the main type of infant feeding (Dubois and Girard, 2003). Since higher socioeconomic status (SES) is associated with lower basal cortisol concentrations (Lupien et al., 2000, Kristenson et al., 2004), it is important that stress studies in the postpartum population take SES indices into account in data analyses.

Other maternity-related variables, such as parity, or the number of children a woman has, have also been suggested to affect the basal activity of cortisol in mothers of young children. Even though psychosocial studies have reported that the source and nature of stressors differ between primiparous and multiparous mothers (Smith, 1989), cortisol reactivity to emotional and stressful events in primiparous and multiparous mothers has not been thoroughly documented (for a review, see Tu et al., 2005a). Compared to first-time mothers, multiparous mothers show reduced arousal to infant cries (Boukydis and Burgess, 1982), suggesting that they might become habituated to child-related stimuli. Alternatively, having multiple children in the household has been associated with reduced steepness in cortisol decline during the day (Adam and Gunnar, 2001). These conflicting results illustrate the lack of consistency over the influence of multiparity on reactivity to daily life challenges in mothers.

Most studies to date have investigated reactivity to acute laboratory-induced stressors in postpartum mothers, in relation to infant feeding choice (Tu et al., 2005a). For instance, breastfeeding has been associated with reduced cardiovascular features (Light et al., 2000, McEwen and Seeman, 1999, Mezzacappa et al., 2005) and blunted HPA responses to a physical stressor (Altemus et al., 1995), but not following a standardized psychological stressor of a speech and arithmetic task (Altemus et al., 2001) in women tested 8–18 weeks after delivery. Importantly, this blunted responsiveness might be a function of the time elapsed between the last feeding period and the beginning of the experimental session (Heinrichs et al., 2001). In addition, anxiety-reducing properties of prolactin and oxytocin, two hormones involved in milk synthesis and milk-ejection might also contribute to this blunted responsiveness (Neumann et al., 1998; Torner and Neumann, 2002; Uvnas-Moberg, 1998).

Basal cortisol secretion in relation to breastfeeding and maternal parity has been less studied. Fleming et al. (1997), using a single measure of salivary cortisol during the first 2 days postpartum in a hospital setting, found that primiparous mothers who showed greater attraction to infant odors had higher cortisol levels. In a similar hospital setting, Bonnin (1992) showed that compared to non-postpartum mothers, basal cortisol levels in new mothers were higher after birth and progressively decreased after day 4 postpartum. In this limited study of 22 subjects, breastfeeding mothers had higher cortisol levels than bottlefeeding mothers, but there was a significant positive correlation between higher morning cortisol levels and more pronounced depressed mood in bottlefeeding mothers. Primiparous mothers had higher morning cortisol levels than their multiparous counterparts, suggesting a potential influence of multiparity on cortisol secretion. Although this study provided some evidence for the role of infant feeding choice, parity and sleep on morning cortisol secretion, it is hampered by several interfering factors that might have influenced the results. For instance, morning cortisol concentrations collected in a hospital setting are known to be higher than home-based samples (Scheer et al., 2002) and thus might not reflect ‘basal’ conditions. This is particularly pertinent since the comparision group of non-postpartum mothers collected their samples at home. Since only a single cortisol sample was collected between 0800 and 0900 h rather than several samples over the entire day, this study provided limited information on changes in amplitude or variability of the diurnal rhythm.

In past years, studies have shown that in addition to inter-individual differences, significant intra-individual differences exist in diurnal cortisol levels measured over the day that are associated with upper respiratory illness symptom susceptibility or symptom expression (Smyth et al., 1997; Ice et al., 2004; Stone et al., 2001). Variability in diurnal cortisol secretion can be detected in healthy individuals when diurnal cortisol is measured over a 2-day period (Smyth et al., 1997) and the slope of the cortisol decline is calculated. As a result, individuals might present either a ‘consistent decline’ (51% of the population), a ‘flat’ cycle (17%), or an ‘inconsistent’ cycle (31%) in cortisol levels on both days. Unfortunately, in Smyth's study, the pattern of the diurnal cortisol cycle was not related to any demographic or psychosocial measures. Of importance is the observation that a marked diurnal rhythm is associated with more positive health outcomes (Heim et al., 2000).

Based on these findings, we investigated the impact of infant feeding choice (breastfeeding vs. bottlefeeding) and maternal parity (primi- vs. multiparous) on the amplitude of basal diurnal cortisol secretion (magnitude of cortisol concentrations) in postpartum women (5–20 weeks postpartum) and non-postpartum mothers. Salivary cortisol samples were collected at home and analysis of the diurnal changes was performed over two non-consecutive days, taking education, income, trait anxiety characteristics, sleep disturbances and salivary estradiol levels into account. In addition to potential group differences in the magnitude of cortisol concentrations at different times during the day, we tested whether higher variability in the diurnal rhythm was present in postpartum compared to non-postpartum mothers, and occurred as a function of feeding choice and parity.