The hypothalamic–pituitary–adrenal (HPA) axis regulates cortisol production and the organism's capacity to respond to stressors and thus maintain homeostasis. From the beginning of the second trimester, the HPA axis functions and organizes in response to the environment [1
]. Secretion of the cortisol is essential for lung maturation at birth, hence, there is a corresponding increase in cortisol levels with increasing gestational age [4
]. One month after birth, full-term healthy infants develop a circadian rhythm of salivary cortisol with increasing morning levels and corresponding decreasing evening levels [7
]. In preterm infants, developmental immaturity and/or the effects of critical illness on adrenal function may result in insufficient cortisol production to maintain homeostasis when exposed to a stressor [8
]. Thus, preterm infants may be able to produce enough cortisol to maintain homeostasis under non-stressful conditions, but insufficient cortisol to respond appropriately when exposed to a stressor. An appropriate cortisol release in the face of a stressor is vital for survival and the lack of such a response increases the risk of morbidity and mortality in preterm infants [1
]. On the other hand, longstanding high levels of cortisol may increase the risk of cognitive and behavioral problems, high blood pressure, and development of the metabolic syndrome [10
The life of the preterm infant is inherently stressful from the moment of birth as the environment in a NICU is usually busy. Several infants are cared for in the same room; sounds and alarms from cardio respiratory monitors, incubators, and mechanical ventilators add to their stress. As a part of their medical care, the preterm infants are not only subjected to many different invasive procedures but also bombarded with stimuli from daily handling procedures (for instance: diaper changing, repositioning, weighing, and personal hygiene care). In a Canadian observational study, Johnston and colleagues found that preterm infants were subjected to a mean of six tissue-damaging and 14 non-tissue-damaging procedures in one week [15
]. Grunau and colleagues have shown that children born very preterm exhibit altered HPA axis functioning at three, eight, and 18 months, and at seven years [16
]. Moreover, they found that elevated salivary cortisol levels in eight and 18-month old preterm infants positively correlated to the amount of pain exposure in the neonatal period [18
]. Compared to full-term infants, preterm infants are at greater risk of short-term consequences of stress (e.g., fluctuations in intracranial blood pressure with an increasing risk for intraventricular hemorrhage [21
]), as well as long-term consequences of stress (e.g., allostatic load and an inability to respond appropriately to a stressor [23
In a review from 2009, it was found that infants aged under three months showed increased cortisol levels from baseline in response to painful interventions (e.g., heel lance and inoculation) as well as routine handling interventions (e.g., physical examination, diaper changing and removal from bath) [24
]. Several studies indicated that in relation to routine handling interventions, preterm infants had symptoms of stress such as increased heart rate, skin conductance, and pain scores, and decreased oxygen saturation [25
]. Pleasant interventions such as getting a massage and listening to recorded sounds of the maternal heartbeat have been shown to decrease salivary cortisol levels in full-term infants [30
]. A combination of sweet-tasting oral solution and pacifier as pain relief during vaccination has also been shown to decrease salivary cortisol levels in full-term infants [32
], as has non-nutritive sucking during circumcision [33
]. Moreover, practicing neonatal care methods such as family-centered care, skin-to-skin contact (SSC) and holding have been proved to have a beneficial impact on the developing HPA axis in preterm infants [34
]. However, no review has summarized preterm infants’ cortisol reactivity in response to painful, handling and pleasant interventions, respectively.
Mode of delivery and prenatal exposure to maternal stress have also been shown to affect the subsequent cortisol reactivity of the infant [37
]. For example, Taylor and colleagues found a lower cortisol reactivity during a four-month-vaccination among infants delivered by cesarean-section compared to vaginally-born infants [37
]. Similarly, O'Connor and collaborators found lower cortisol reactivity during stress provocation (Ainsworth’s strange situation) in 17-month-old infants who were exposed to high levels of cortisol in utero
It is important to know what stressors are suitable to use to evaluate salivary cortisol reactivity in preterm infants in order to guide researchers in future studies. Available reviews were focused on salivary cortisol reactivity in response to acute stressors in adults [40
] and children [24
]. Cortisol in saliva has been used as a valid method to measure adrenocortical activity in newborns since 1987 [42
]; however, no reviews have been conducted to summarize the salivary cortisol reactivity in relation to acute stressors in preterm infants in the NICU. Moreover, it is important to evaluate the type of interventions performed in the NICU that are beneficial for preterm infants in order to prevent longstanding high levels of cortisol and allostatic load.
The aim of this review was to summarize the evidence from interventions leading to a change in salivary cortisol levels from the baseline (before the intervention) to response (after the intervention) in preterm infants in the NICU. The specific research questions were:
What interventions lead to an increase in salivary cortisol levels from the baseline to response in preterm infants in the NICU?
What interventions lead to a decrease in salivary cortisol levels from the baseline to response in preterm infants in the NICU?
This review summarizes the evidence of painful, handling, or pleasant interventions that caused a change in salivary cortisol from baseline in preterm infants in the NICU.
Two studies reported significantly increased salivary cortisol levels among preterm infants after exposure to a painful procedure [49
]. These two studies, which recruited preterm infants born at 23–37 weeks GA, reported a successful saliva sampling above 75% and showed significant results despite relatively small sample sizes [49
]. The result is congruent with previous studies, which found that healthy infants usually show an elevation in cortisol in response to painful procedures [24
]. A reaction to pain is a defense mechanism aiming to protect the person from harm and to maintain homeostasis, and strong stimuli such as pain tends to yield non-habituation in healthy infants [61
It is well known that pain relief such as sweet-tasting oral solutions and non-nutritive sucking have proven to be effective in reducing pain during single painful procedures in newborns [63
]. The administration of optimal pain relief probably explains the absence of salivary cortisol reactivity in response to the heel lance in the study by Cignacco et al.
]. In the study of Campbell-Yeo et al.
], infants were given oral sucrose and a pacifier before the heel lance, and co-bedded twins experienced a decrease in salivary cortisol levels, but not separated twins. A dampened cortisol response in infants receiving a pacifier and a sweet-tasting oral solution during a painful procedure is likely to reflect an absence or a lower degree of pain and stress. A significant decrease in response to the heel lance without any pain relief as with the infants exposed to antenatal betamethasone in the study by Davis et al.
] is more likely to be an effect of suppressed adrenal activity due to the corticosteroid use [67
], which is not the same as no pain or stress. It is also possible that co-bedded twins were experiencing relief due to the reunion, which affected the adrenocortical response through human touch and closeness [34
]. Thus, it is possible that the cortisol level decrease was exclusively due to co-bedding the twins. However, it is noteworthy that the nurses who performed the heel lances in the study by Campbell-Yeo et al.
were not blinded to group assignment and the successful saliva sampling was rather low [48
]. Badiee et al.
revealed similar results. They found a significantly higher cortisol level after the heel lance in the standard care group compared to the co-bedding group, but did not provide cortisol reactivity results [47
]. However, the results are promising and, therefore, the replicability of these findings needs to be further investigated. Kleberg et al.
used behavioral support but found no effect of NIDCAP on cortisol levels during the eye-screening examination. However, they found a more rapid recovery in the NIDCAP group [49
]. Recovery, the degree to which cortisol elevations persist after termination of the stressor [69
], was studied in only two of the included articles [49
]. Since preterm infants are subjected to several procedures each day, it is important to continue to investigate the recovery from stressful stimuli in preterm infants in order to learn more about their stress system development and to be able to provide optimal support in the NICU.
Few reviewed studies used handling as a stressor [29
]. Magnano and colleagues found that healthy preterm infants did not respond to a physical examination while antenatal cocaine-exposed infants’ salivary cortisol levels increased [52
]. This result is in accordance with the theory of habituation. Repeated milder stressors that do not involve a threat to physical or psychological well-being may yield habituation of the adrenocortical response in healthy infants but not necessarily in unhealthy infants [61
It is of the utmost importance to find care methods that can buffer stress and, thus, longstanding elevated levels of cortisol in preterm infants in the NICU. Two studies found decreased salivary cortisol levels in response to a changed positioning from lateral/supine to prone and live harp music, respectively [57
]. However, both studies used the single group method, which makes it difficult to determine whether the decrease in cortisol levels was due to the intervention or a natural decline in cortisol levels [7
]. Moreover, coefficients of variation for the assay methods were not reported, which affects the reliability. Since the results are important and interesting, both studies should be repeated using a randomized, controlled study design.
An absence of cortisol response in a stressful situation which may threaten the homeostasis of the premature infant may have several causes. First, suppressed adrenal activity due to corticosteroid treatment [67
] or longstanding high stress load leading to allostatic load and an inability to respond [23
]. Second, a high baseline value, resulting in a smaller rise as explained by the law of initial value [70
]. Third, immaturity of the HPA axis system causing inconsistent, less robust cortisol responses [4
]. Fourth, the stressor is too mild and/or the infant has become habituated to the procedure [62
]. Fifth, methodological issues such as study design, lack of power, saliva collection time, and quality of the cortisol analysis method.
In this review, several studies reported an absence of cortisol response in relation to painful procedures [44
], handling [29
], and pleasant interventions [56
]. Boyer et al.
presented some methodological issues with low successful saliva sampling and no report of analysis method [51
]. Dorn et al.
] reported a rather low rate of successful saliva sampling, no intra-assay coefficient of variation, and moreover collected response values after 10 min instead of the recommended 20–30 min [59
]. In studies by Ivars et al.
and Mörelius et al.
2006 and 2012, it is plausible that the stressors were too mild for rather stable infants, and so the studies ought to be repeated with more well-defined groups comprising only infants with low GA and low postnatal age [29
]. If the parent is performing the diaper change, it is also important to control for parental sensitivity and SSC, since both parental sensitivity and SSC may buffer the infant’s stress reaction [34
]. In 2005, Mörelius et al.
included infants with a wide range of GA from 25 to 33 weeks during SSC [56
]. The results show a large variability in the cortisol responses with both elevated and decreased levels reflecting an immature HPA axis. Several studies included in this review examined an unselected group of preterm infants in terms of GA. Extremely preterm infants are more immature and there is no correlation between ACTH and cortisol as there is in more mature infants [19
]. Hence, researchers should consider the GA of preterm infants when designing studies of salivary cortisol reactivity in the future, and should aim at creating more homogenous groups for better comparison.
Other methodological issues such as obtaining optimal sample volumes of saliva without using a saliva stimulant could be reasons for the small number of studies performed in the NICU. A sampling time exceeding 5 min [57
], cheek massage [48
], or aspiration of saliva with a syringe [53
] may be stressors in themselves and may confound the cortisol results. It is therefore valuable to adopt non-invasive sampling methods as well as measurement methods using minute sample volumes when studying salivary cortisol in preterm infants [72
In this review, we have summarized the cortisol reactivity in relation to different interventions but we have not evaluated the interventions per se. Thus, it is possible that the included interventions, e.g., diaper change and SSC, may still be stressful or have a calming effect for the preterm infants in the NICU even though there was no change in cortisol levels. Studies that did not report ethical approval were also included in this review and the advisability of such inclusion may be debatable. However, the reasons for inclusion of these studies were that both studies used stressors that were part of the infants’ care and the results are valuable for the understanding of stress reactivity in preterm infants.