Environmental stress and salivary cortisol levels in preterm infants

Abstract

BACKGROUND:

Preterm infants are exposed to numerous environmental stressors during their Neonatal Intensive Care Unit (NICU) stay, particularly during the first week after birth. The aim of this study is to assess whether salivary cortisol levels are correlated with Neonatal Infant Stressor Scale (NISS) scores in preterm infants during the first week of life. We also quantified the changes in both NISS scores and cortisol levels in the first week, and whether cortisol levels are associated with gestational age.

METHODS:

Preterm infants (n = 38, birth weight <1250 g and/or gestational age <29 weeks) were included. Saliva samples were collected on day 0–3 (early) and day 4–7 (late), and cortisol concentrations were measured by immunoassay. NISS scores were assessed retrospectively for the six hours preceding each saliva collection.

RESULTS:

NISS scores were not significantly correlated with salivary cortisol levels at either time point. However, infants born at <28 weeks gestation had higher median cortisol levels than infants born at >28 weeks (p = 0.0068), and there was also a significant inverse relationship between NISS score and gestational age (p = 0.04). There was no significant difference between the early and late time points for either NISS scores or cortisol levels.

CONCLUSIONS:

Cortisol levels are elevated in infants <28 weeks gestation but do not correlate with NISS scores. NISS scores are inversely related to gestational age, likely reflecting increased exposure to interventions and invasive procedures for the smallest infants.

Keywords

Environmental stress NISS score preterm infants salivary cortisol

1 Introduction

Preterm infants are exposed to numerous stressors during their NICU stay, particularly during the first week after birth. While infants are undergoing physiologic changes of postnatal adaptation, the NICU environment introduces stressors including noise, light, temperature fluctuations, and interventions such as line placement, imaging, blood draws, and diaper changes. Excessive stress or agitation can contribute to adverse outcomes such as intraventricular hemorrhage (IVH), which can lead to long-term physical and developmental disability. IVH occurs because extremely preterm infants are born with fragile blood vessels in the periventricular germinal matrix. Even small fluctuations in blood pressure induced by stress can lead to bleeding and trigger neuronal injury or ventricular dilation. Almost all IVH occurs during the first week of life, so minimizing stress during that time is an important target for improving the quality of NICU care and outcomes [1].

To minimize stress in the NICU, clinicians need tools to measure it. The Neonatal Infant Stressor Scale (NISS) is a composite score based on the number and intensity of hands-on events and interventions that are associated with NICU-induced stress, such as mechanical ventilation, blood draws, line placement, IV flushes, positioning, diaper changes, and suctioning. The NISS score was developed in 2009 by Newnham, et al. [2] and validated in a cohort study with a NICU population 28–32 weeks gestation [3]. However, it requires extensive documentation and is difficult to apply universally in the clinical setting.

Alternatively, stress can be quantified using physiologic biomarkers, including cortisol and amylase. Cortisol is released via the hypothalamic-pituitary-adrenal (HPA) axis in response to physical or emotional stressors. In infants, cortisol levels rise after painful procedures or noxious stimuli and decrease after soothing interventions [4 –6]. In preterm infants, salivary cortisol levels correlate well with serum cortisol levels and increase acutely in response to pain or stress [7 –9]. Newborns do not establish diurnal variation in cortisol levels during the first weeks of life, allowing for a more reliable comparison among samples and eliminating the potential confounder of time in interpreting responses [8, 10]. Alpha amylase, an enzyme in saliva, also increases in response to adrenergic stimuli, including acute stressors [11 –13]. Concentrations of both cortisol and alpha amylase in saliva can be measured using commercially available assays, and saliva can be collected from preterm infants non-invasively using established methods [3, 14].

Although previous studies have established a relationship between stressors and salivary biomarkers in larger and older infants, the utility of salivary cortisol and amylase for assessing stress in extremely preterm infants in the first week of life is not known. Validation of non-invasive biomarkers of stress could support efforts to identify and minimize harmful exposures for these infants during their most vulnerable period. The primary aim of this study was to assess whether salivary cortisol levels are correlated with Neonatal Infant Stressor Scale (NISS) scores in extremely preterm infants during the first week of life.

2 Materials & methods

This is a prospective observational study enrolling preterm infants born at Cohen Children’s Medical Center (Level IV NICU) or North Shore University Hospital (Level III NICU) between October 2021 and February 2023. The study was approved by the Institutional Review Board of Northwell Health (#21-0773), and informed consent was obtained from the mother of each participant.

2.1 Subjects

Infants were eligible for the study if they had birth weight ≤1250 g and/or gestational age <29 weeks. Exclusion criteria included hemodynamic instability, severe IVH (grade three or four), necrotizing enterocolitis (NEC) or intestinal perforation, acute distress, or were otherwise deemed medically unstable by the primary team. Additionally, the infant’s mother needed to be able to provide consent for enrollment within the infant’s first three days of life.

2.2 Clinical characteristics

Clinical and demographic data were obtained from the electronic medical record, including: (1) maternal demographics and antenatal course (e.g., maternal substance use, pregnancy complications, antenatal steroid exposure, and singleton vs. multiple gestation), (2) delivery information and delivery room course (e.g., gestational age, birthweight, APGAR scores, and level of resuscitation required), and (3) neonatal complications experienced either during the study period or the remainder of the NICU stay, including hypoglycemia.

2.3 Saliva specimen collection and analysis

Two saliva samples were collected for each subject: an “early” sample collected between day of life (“DOL”) zero and three, and a “late” sample collected between DOL four and seven. Saliva was collected using the Salimetrics SalivaBio Infant Swab. One end of the swab was placed in the infant’s cheek for approximately five minutes or until saturated with saliva. Safety monitoring data was documented at the time of specimen collection, including heart rate, respiratory rate, and oxygen saturation level both at baseline and during each minute of collection. Collection was stopped immediately if the infant experienced significant changes in any of these biometric parameters.

Immediately after saliva collection, the swab was placed inside the top portion of a Salimetrics double-chambered collection tube and centrifuged. The extracted saliva was stored at –60°C until batched analysis (Dresden Lab Service GmbH, Dresden, Germany). Cortisol and alpha amylase levels were measured using chemiluminescence immunoassays with high sensitivity (Tecan - IBL International, Hamburg, Germany; catalogue number R62111). The intra- and interassay coefficients of variance were below 9%.

2.4 Neonatal Infant Stressor Scale (“NISS”) assessment

Investigators used data from the electronic health record and interviewed each subject’s bedside nurse to generate a NISS score reflective of the six hours prior to each saliva collection. For NISS scoring, each care event was assigned a numerical point value from one to five, with more stressful events receiving a higher point value. The total number of points generated by all care events during the six hours preceding saliva collection was tallied to yield a single numerical composite score. A maximum theoretical score of 107 was established after exclusion of items in the NISS tool that are not applicable to this population of preterm infants in the first week of life, such as bottle feeding and eye examination.

2.5 Statistical analysis

Descriptive statistics (e.g., frequency and proportion for categorical variables and means and standard deviations or median and interquartile range for continuous variables) were computed. Due to the distribution of NISS scores in the sample at both time points, NISS scores for both time points were assessed as both a continuous variable and categorical for the primary aim. NISS scores were categorized into quartiles for each sample. For the first time point, NISS was categorized as: 7–17, 19–21, 22–27 and 28–34. At the second time point, NISS was categorized as: 12–18, 19–21, 22–25 and 26–41. Due to the limited sample size, Monte Carlo simulation for exact tests were utilized, as appropriate. Due to the distribution of cortisol levels in the sample at both time points, non-parametric tests were utilized for all relevant analyses. Specifically, for the primary aim, exact Kruskal-Wallis tests were used to compare the distribution of cortisol levels between the NISS quartiles as well as to evaluate the association between NISS quartiles and gestational age as a continuous variable. Monte Carlo estimation for the exact Wilcoxon rank sum test was utilized to assess the relationship between cortisol levels and gestational age (<28 vs ≥28 weeks) and between cortisol levels and hypoglycemia. The association between hypoglycemia and gestational age was assessed using a Chi-square test. The correlation of salivary cortisol levels with NISS scores (analyzed continuously) was evaluated using Spearman rank correlation.

To determine the change in cortisol levels in the first week of life, cortisol levels taken at the first time point were compared to levels at the second time point using Wilcoxon signed rank test for matched pairs. The same test was used to measure the change in NISS score in the first week of life (difference in NISS scores between first and second time point). The Fisher’s exact test was used when assessing the relationship between gestational age as a binary variable (<28 vs ≥28 weeks) and NISS quartiles (first sample), and between antenatal steroid use group (≤24 hours, >24 hours to≤one week, ≥ one week) and NISS quartiles. The level of significance was set at 5%. All aims were analyzed using SAS software Version 9.4 (Cary, NC).

3 Results

3.1 Demographics

A total of 38 patients meeting inclusion criteria were enrolled in this study, with mean±standard deviation (SD) birth weight of 957±215.6 g and mean gestational age of 28.1±2.4 weeks. Subject demographics and clinical characteristics are summarized in Table 1. Of the 38 enrolled subjects, 35 had “early” (DOL 0–3) and 32 had “late” (DOL 4–7) saliva samples available for analysis (Fig. 1). Attrition was due to factors such as the development of exclusion criteria (severe IVH or intestinal perforation), withdrawal from the study, death, receipt of exogenous steroids, or technical issues relating to sample collection (e.g., insufficient volume of saliva for analysis).

Table 1
Baseline characteristics of study participants (n = 38)

Characteristic Value^*

Birth weight (g) 957±216

Gestational age (weeks) 28.1±2.4

Gender

Male 21 (55%)

Female 17 (45%)

APGAR score

1 minute 5.1±2.8

5 minutes 7.4±1.7

Antenatal steroids

None 1 (3%)

<24 h prior to delivery 10 (26%)

1–7 days prior to delivery 17 (45%)

>7 days prior to delivery 10 (26%)

Pregnancy complications

IUGR 11 (29%)

Preterm labor 18 (47%)

PPROM 12 (32%)

Gestational diabetes 6 (16%)

PEC/gHTN 14 (37%)

Delivery room resuscitation

CPAP 14 (37%)

PPV 16 (42%)

Intubation 7 (18%)

Chest compressions 1 (3%)

Characteristic	Value^*
Birth weight (g)	957±216
Gestational age (weeks)	28.1±2.4
Gender
Male	21 (55%)
Female	17 (45%)
APGAR score
1 minute	5.1±2.8
5 minutes	7.4±1.7
Antenatal steroids
None	1 (3%)
<24 h prior to delivery	10 (26%)
1–7 days prior to delivery	17 (45%)
>7 days prior to delivery	10 (26%)
Pregnancy complications
IUGR	11 (29%)
Preterm labor	18 (47%)
PPROM	12 (32%)
Gestational diabetes	6 (16%)
PEC/gHTN	14 (37%)
Delivery room resuscitation
CPAP	14 (37%)
PPV	16 (42%)
Intubation	7 (18%)
Chest compressions	1 (3%)

^*Value indicates n (%) or mean±standard deviation. IUGR, intrauterine growth restriction; PPROM, preterm premature rupture of membranes; PEC, pre-eclampsia; gHTN, gestational hypertension; CPAP, continuous positive airway pressure; PPV, positive pressure ventilation.

Fig. 1

Flow diagram of study participants.

3.2 NISS scores

The mean±SD NISS scores were 21.7±6.6 “early” and 22±6.2 “late” (p = 0.65). However, when NISS score was assessed using quartiles, the distribution of gestational age was significantly different among the “early” scores. Specifically, higher gestational age was seen among the first and second quartiles compared with third and fourth quartiles (p = 0.04, Fig. 2).

Fig. 2

Distribution of NISS scores by gestational age. Boxplot showing gestational age (continuous variable) by NISS score quartile at the “early” time point. Higher gestational age was seen among the first and second quartiles compared with the third and fourth quartiles (p = 0.04).

3.3 Salivary biomarkers

The median (Q1–Q3) of salivary cortisol levels was 10.3 (6.2 to 19.7) nmol/L for “early” samples and 10.7 (5.6 to 28.1) nmol/L for “late” levels (p = 0.37). Time since antenatal maternal steroid use (≤24 hrs, >24 hours to 1 week, ≥1 week) was not related to “early” neonatal cortisol levels (p = 0.36). The distribution of cortisol levels was significantly higher in infants <28 weeks gestation [16.6 (8.1 to 32.5) nmol/L] compared with infants >28 weeks gestation [9.2 (4.8–12.4) nmol/L] (p = 0.0068, Fig. 3). Hypoglycemia occurred in 39% of infants <28 weeks gestation and in 60% of infants >28 weeks gestation (p = 0.19). Cortisol levels were not significantly different between hypoglycemic and normoglycemic infants at either time point (p = 0.31 “early,” p = 0.27 “late”). Salivary amylase was undetectable in 44 of 48 samples assayed, so amylase levels were not included in further analyses.

Fig. 3

Distribution of cortisol levels by gestational age. Boxplot showing the distribution of cortisol levels (nmol/L) by gestational age (<28 weeks vs. ≥28 weeks) at the “early” time point (p = 0.0068).

3.4 Correlation of salivary cortisol with NISS scores

“Early” salivary cortisol levels were not significantly correlated with NISS scores analyzed either continuously (p = 0.27) or by quartiles (p = 0.82). Similarly, “late” cortisol levels did not correlate with NISS scores (p = 0.21 and p = 0.56, respectively).

4 Discussion

Cortisol levels are higher in infants <28 weeks than in infants ≥28 weeks gestation but are not directly related to stressful interventions in the NICU that are quantified by the NISS score. Our finding that the smallest and most vulnerable infants have the most robust cortisol responses indicates that they experience the highest levels of physiologic stress, even though environmental exposures are not the primary drivers of that stress. Instead, it is likely that the cortisol response in extremely premature infants in the first week is driven by the stress of prematurity and medical instability per se. Extremely premature infants are most susceptible to morbidities associated with stress, such as IVH, so NICU practices should aim to minimize agitation. However, our findings suggest that the primary sources of physiologic stress for extremely preterm infants are not readily characterized or isolated. It remains unclear how to define environmental stress in NICU and whether it is preventable.

It is important to note that NISS scores were not normally distributed and had limited variability. Although the maximum theoretical score was 107 in our population, the mean was 21.7 with a standard deviation of 6.6 (IQR 19–28). This is because many of the infants in the study had a very similar early clinical course, and all were being cared for according to a “small baby care bundle” that is based on principles of minimal stimulation. The typical infant enrolled in this study was receiving CPAP and orogastric tube feeds with regular diaper changes and the occasional suctioning or telemetry lead replacement. Our findings suggest that the NISS score may not be a sensitive tool for assessing stressful events among premature infants receiving contemporary NICU care during the first week, but further research involving a larger sample size should be conducted to investigate this further.

One of the limitations of this study is that there are many confounding factors that may have an impact on the infant’s cortisol response, including the intrauterine milieu, maternal cortisol level, infection, and maturity of the hypothalamic-pituitary-adrenal axis. Conversely, our findings support previous reports that hypoglycemia is not directly related to cortisol levels in newborns, as would be expected in older children and adults [15, 16]. The reference ranges for cortisol in preterm neonates are wide, particularly in the first weeks of life, with random measurements reported up to >500 nmol/L [17, 18]. Therefore, “normal” cortisol responses are not well defined in the “early” or “late” periods in the first week of life for preterm infants.

While we tried to capture as many clinically relevant variables as possible, the small sample size did not allow us to adequately adjust for confounders as in a multiple linear regression model. Similarly, the comparison of “early” to “late” NISS scores and cortisol levels did not consider the time between measurements (between one and seven days), which may have impaired detection of time-dependent changes. It is also possible that this study was inadequately powered to detect a statistically significant direct correlation between NISS scores and cortisol levels. Barriers to enrollment included parental hesitation and inability to obtain consent prior to DOL three due to maternal medical status.

We could not assess the potential relationship between alpha amylase and environmental stress because we could not detect it in most samples. This is likely due to low physiologic levels of salivary amylase expression in newborns. Previous reports have suggested that salivary alpha amylase increases with gestational age and throughout the first several months of life. Therefore, although amylase responses are proportionate to stress in adults, this has not been confirmed in neonates [13 , 19–23].

In conclusion, infants born at <28 weeks have higher NISS scores and higher salivary cortisol than infants born at ≥28 weeks gestational age. However, NISS scores and salivary cortisol levels are not directly correlated in preterm infants during the first week. Ongoing research is needed to define, quantify, and prevent environmental and iatrogenic stress in the NICU.

Disclosures

The authors have no conflicts of interest to disclose. The project had no external funding/support. The project was approved by the Institutional Review Board of Northwell Health (#21-0773).

References

Aucott

, Watterberg

, Shaffer

, Donohue

; PROPHET study group. Early cortisol values and long-term outcomes in extremely low birth weight infants. J Perinatol. 2010;30(7):484–8.

Newnham

, Inder

, Milgrom

. Measuring preterm cumulative stressors within the NICU: The neonatal infant stressor scale. Early Hum Dev. 2009;85(9):549–55.

Pourkaviani

, Zhang

, Spear

, D’Agostino

, Satty

, Liu

, et al. Clinical validation of the neonatal infant stressor scale with preterm infant salivary cortisol. Pediatr Res. 2020;87(7):1237–43.

Kleberg

, Warren

, Norman

, Mörelius

, Berg

, Mat-Ali

, et al. Lower stress responses after newborn individualized developmental care and assessment program care during eye screening examinations for retinopathy of prematurity: A randomized study. Pediatrics. 2008;121(5):e1267–78.

De Bernardo

, Riccitelli

, Giordano

, Proietti

, Sordino

, Longini

, et al. Rooming-in reduces salivary cortisol level of newborn. Mediators Inflamm. 2018;2018:2845352.

Mörelius

, He

, Shorey

. Salivary cortisol reactivity in preterm infants in neonatal intensive care: An integrative review. Int J Environ Res Public Health. 2016;13(3):337.

Chou

, Lien

, Lin

, Fu

, Kao

, Tsai

, et al. The relationship of salivary and cord blood cortisol in preterm infants. J Pediatr Endocrinol Metab. 2011;24(1-2):85–8.

Castro

, Elias

, Martinelli

Jr , Antonini

, Santiago

, Moreira

. Salivary cortisol as a tool for physiological studies and diagnostic strategies. Braz J Med Biol Res. 2000;33(10):1171–5.

Forclaz

, Moratto

, Pennisi

, Falco

, Olsen

, Rodríguez

, et al. Salivary and serum cortisol levels in newborn infants. Arch Argent Pediatr. 2017;115(3):262–6. English, Spanish.

10.

Ivars

, Nelson

, Theodorsson

, Ström

, Mörelius

. Development of salivary cortisol circadian rhythm in preterm infants. PLoS One. 2017;12(8):e0182685.

11.

Nater

, Rohleder

. Salivary alpha-amylase as a non-invasive biomarker for the sympathetic nervous system: Current state of research. Psychoneuroendocrinology. 2009;34(4):486–96.

12.

Steckl

, Ray

. Stress biomarkers in biological fluids and their point-of-use detection. ACS Sens. 2018;3(10):2025–44.

13.

Ali

, Nater

. Salivary alpha-amylase as a biomarker of stress in behavioral medicine. Int J Behav Med. 2020;27(3):337–42.

14.

Tryphonopoulos

, Letourneau

, Azar

. Approaches to salivary cortisol collection and analysis in infants. Biol Res Nurs. 2014;16(4):398–408.

15.

Crofton

, Midgley

. Cortisol and growth hormone responses to spontaneous hypoglycaemia in infants and children. Arch Dis Child. 2004;89(5):472–8.

16.

Jackson

, Williams

, Burchell

, Coughtrie

, Hume

. Plasma catecholamines and the counterregulatory responses to hypoglycemia in infants: A critical role for epinephrine and cortisol. J Clin Endocrinol Metab. 2004;89(12):6251–6.

17.

Heckmann

, Wudy

, Haack

, Pohlandt

. Reference range for serum cortisol in well preterm infants. Arch Dis Child Fetal Neonatal Ed. 1999;81(3):F171–4.

18.

. Is there a “normal” range of serum cortisol concentration for preterm infants? Pediatrics. 2008;122(4):873–5.

19.

Davis

, Granger

. Developmental differences in infant salivary alpha-amylase and cortisol responses to stress. Psychoneuroendocrinology. 2009;34(6):795–804.

20.

Bellavia

, Moreno

, Sanz

, Picas

, Blanco

. Alpha-amylase activity of human neonate and adult saliva. Arch Oral Biol. 1979;24(2):117–21.

21.

Ben-Aryeh

, Lapid

, Szargel

, Benderly

, Gutman

. Composition of whole unstimulated saliva of human infants. Arch Oral Biol. 1984;29(5):357–62.

22.

Hodge

, Lebenthal

, Lee

, Topper

. Amylase in the saliva and in the gastric aspirates of premature infants: Its potential role in glucose polymer hydrolysis. Pediatr Res. 1983;17(12):998–1001.

23.

Sevenhuysen

, Holodinsky

, Dawes

. Development of salivary alpha-amylase in infants from birth to 5 months. Am J Clin Nutr. 1984;39(4):584–8.