Report of Original Research
Effects of Antenatal Corticosteroids on Cortisol and Heart Rate Reactivity of Preterm Infants
Biological Research for Nursing 1-8 ª The Author(s) 2015 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/1099800414564860 brn.sagepub.com
Sandra J. Weiss, PhD, DNSc, RN, FAAN1, and Sandra Niemann, PhD, RN1
Abstract Administration of glucocorticoids (GCs) during pregnancy is an established practice for reducing morbidity and mortality of fetuses at risk of preterm delivery. However, preliminary research indicates that exposure to exogenous GCs in utero may be associated with suppressed hypothalamic–pituitary–adrenal axis activity. The aim of this study was to determine whether preterm neonates who are exposed to antenatal corticosteroids show evidence of a suppressed stress-response system during their first few weeks of life, in contrast to infants who are not exposed. The sample (51% female) included 57 neonates, with 74% exposed to steroids. Mean gestational ages (GAs) were 32.6 weeks for exposed and 34.7 weeks for nonexposed infants. Although neonates in the two groups differed in gender, birth weight, and morbidity, these factors were controlled for in data analyses. Infants’ salivary cortisol and heart rate (HR) were measured before and after they received a standardized caregiving ‘‘stressor’’ while in the newborn intensive care unit. Infants exposed to GCs in utero had lower basal cortisol levels and higher HRs than their nonexposed peers. In contrast to infants who received no GCs, they also exhibited minimal HR or cortisol reactivity to the stressor. Findings suggest that preterm infants who were exposed to antenatal corticosteroids experience a suppressed response to stress. As preterm children develop, this dysregulation has numerous implications for later development of stress-related cardiovascular and mental health problems. Further research is needed to determine whether these postnatal effects of antenatal corticosteroids persist over time. Keywords corticosteroids, HPA axis, infant stress response, prematurity
Preterm birth is a major health issue for families and for the health care system. Approximately 12% of the U.S. births are preterm (Martin et al., 2012). Prematurity represents a risk factor for biobehavioral and neurodevelopmental disorders that lead to multiple health problems in childhood and adulthood (Gluckman & Hansen, 2004; Gonc¸alves, Chaiworapongsa, & Romero, 2002; Thompson, Syddall, Rodin, Osmond, & Barker, 2001). Infant morbidity associated with prematurity clearly contributes to these problems; however, well-meaning treatments may inadvertently influence these outcomes as well. For example, it has become the standard of care in obstetric practice to give antenatal corticosteroids to women at risk of preterm birth (Gilstrap et al., 1995; Roberts & Dalziel, 2006). Although the short-term benefits of this treatment are widely recognized, recent findings raise concerns regarding its effects on development of the fetal hypothalamic–pituitary–adrenal (HPA) axis and autonomic nervous system (ANS), interrelated structures that play key roles in determining resilience or vulnerability to stress. Administration of betamethasone or dexamethasone, the two most commonly used antenatal corticosteroids, is designed
to mimic the rise in cortisol that occurs shortly before birth (Rog-Zielinska, Richardson, Denvir, & Chapman, 2014). As synthetic corticosteroids, however, they have somewhat different properties than endogenously produced glucocorticoids (GCs). Most importantly, synthetic corticosteroids are not deactivated by the placental enzyme 11B-HSD2 (Alexander et al., 2012), thus putting the fetal brain at risk for substantial exposure to the drugs. Recent research using rigorously controlled animal models has raised concerns about the impact of exposure to antenatal corticosteroids on development of the HPA axis. Exposure to antenatal steroids has been associated with attenuated fetal GC secretion, blunted function of the HPA
1
Department of Community Health Systems, School of Nursing, University of California, San Francisco, CA, USA Corresponding Author: Sandra Weiss, PhD, DNSc, RN, Department of Community Health Systems, School of Nursing, University of California, Box 0608, 2 Koret Way, San Francisco, CA 94143, USA. Email: [email protected]
Downloaded from brn.sagepub.com at UNIV NEBRASKA LIBRARIES on September 29, 2015
2
Biological Research for Nursing
axis, and permanent alteration of HPA-axis reactivity later in life (Burlet et al., 2005; Kapoor, Petropoulos, & Mathews, 2008; Shoener, Baig, & Page, 2006). Although studies are few, research indicates that human infants are also affected by synthetic GCs. For term infants, stress-induced responses measured at 24 hr postnatal (Davis, Waffarn, & Sandman, 2011) indicated larger cortisol reactivity in infants exposed to GCs than in infants not exposed. In four studies of preterm infants exposed to GCs, investigators examined infants’ stress responses at 3–6 days postnatal (Ashwood et al., 2006; Davis et al., 2004, 2006; Schaffer, Luzi, Burkhardt, Rauh, & Beinder, 2009), with Davis et al.’s (2006) study also assessing responses during the infants’ second week postnatal. Results for both the Davis et al. and Schaffer et al. studies indicated a suppressed cortisol response in the preterm infants exposed to GCs, in contrast to preterm infants who were not exposed. Ashwood et al. found that 38% of preterm infants receiving a single course of GCs had a suppressed cortisol response, while 68% of infants receiving a repeat course showed suppression. The only other study of which we are aware found that antenatal GCs predicted a suppressed cortisol response at 4 months postnatal in response to immunization (Glover, Miles, Matta, Modi, & Stevenson, 2005). These consistent findings of suppression among GC-exposed preterm infants could stem from enhanced negative feedback on the HPA axis as a result of fetal GC exposure and/or increased tissue sensitivity to GCs (Harris & Seckl, 2011). The HPA axis acts in close concert with the ANS in response to stress (Laurent, Powers, & Granger, 2013). Both systems are regulated by neurons in the paraventricular nucleus of the hypothalamus and have bidirectional effects (Ulrich-Lai & Herman, 2009). The sympathetic nervous system directly innervates the adrenal cortex, which controls synthesis and release of GC hormones. Circulating GCs, in turn, can also potentiate sympathetic nervous system changes, including increased heart rate (HR). Thus, administration of antenatal corticosteroids may also affect key autonomic indicators such as HR. Only a few studies have examined these immediate or longterm effects, however. Davis et al. (2004) reported that HR increased significantly (p ¼ .0001) in response to a heel-stick stressor in steroid-exposed infants compared to controls at 3–6 days postnatal, although the overall HR did not differ between groups. These same infants had a suppressed cortisol response to the heel-stick stressor, suggesting that antenatal steroids may exert opposite early effects on the neuroendocrine and autonomic systems. Other researchers who investigated longer term effects of fetal exposure to steroids found no significant differences in HR between treatment groups for children 7–10 years of age (de Vries et al., 2008) or 30-yearold adults (Dalziel et al., 2005). However, neither of these studies examined HR reactivity to stress. This pilot study was conducted to determine the feasibility of data collection procedures and initial effect sizes for estimating sample size needs of a larger study. It also adds to our knowledge about the effects of antenatal corticosteroids on the HPA axis and ANS in the immediate postnatal period. The aim
of this research was to determine whether preterm infants who are exposed to corticosteroids during gestation show a suppressed stress-response system during the first few weeks of life, as evidenced by dampened cortisol response and HR reactivity. Our hypothesis was that infants exposed to corticosteroids during gestation will have less cortisol reactivity and less HR reactivity in response to a stressor than infants who are not exposed to corticosteroids. Although our hypothesis regarding cortisol is congruent with most previous research, our hypothesis related to HR is not informed by the one previous study showing increased HR for steroid-exposed infants in response to a stressor. Instead, our HR hypothesis was based on previous research regarding the strong, positive relationship between cortisol and HR reactivity in preterm infants (Haley, Grunau, Weinberg, Keidar, & Oberlander, 2010) and the established interconnectedness of the two regulatory systems (e.g., Janig, 2006; Stalder, Evans, Hucklebridge, & Clow, 2011). Thus, we expected to see a parallel coupling of physiologic responses and a general pattern of suppression across both autonomic and neuroendocrine systems.
Method Procedures Infants were eligible to participate if they were born between 28 and 37 weeks’ gestation and had no known brain malformation, genetic anomaly, or risk of dying. In the obstetrics service at our institution, between 70% and 80% of women who deliver preterm infants are given antenatal corticosteroids by their clinicians. As a result, we did not expect the sample to be evenly distributed between groups of infants who were exposed to corticosteroids and those who were not. Research nurses in the Pediatric Clinical Research Center identified eligible infants in the newborn intensive care unit (NICU) and provided a flyer about the study to their mothers. Mothers who indicated an interest in participating were approached by the research team to obtain written consent. This study was approved by the Institutional Review Board of the university. We assessed each infant’s stress response (cortisol and HR reactivity) twice in the NICU 2–4 weeks after birth. The stressor included 10 min of care provided by a neonatal nurse. Caregiving included changing the infant’s diaper, performing peri care, taking the temperature, checking the feeding tube, and measuring abdominal girth. The protocol for the caregiving stressor was based upon standard care provided to infants in the NICU and developed in consultation with the NICU nurses. This process assured that all nurses performing the care would be familiar with the protocol and comfortable with its safety for the infants. We held a meeting with the nurses at the beginning of the study to review the protocol and answer any questions about its implementation. To enhance fidelity, we gave the nurse providing the care a small card to review at the start of each data collection session that highlighted the components of the protocol and the order in which they should occur. The card remained near the crib or incubator during the entire
Downloaded from brn.sagepub.com at UNIV NEBRASKA LIBRARIES on September 29, 2015
Weiss and Niemann
3
procedure for reference, if needed. All caregiving sessions were scheduled between 2 and 4 p.m. in the afternoon, one of the quietest times in the NICU when procedures, interventions, and visitors are minimal. The infant rested without any contact for 15 min prior to data collection. If the infant was not in a quiet state, baseline time was extended to assure a 15-min period at rest. A research assistant (RA) then collected baseline HR data for 5 min before caregiving began. After caregiving, the infant was swaddled and allowed to rest quietly for a 5-min postcaregiving period. The RA again collected HR data during this time. At the end of the baseline period and again after the postcaregiving period ended, the nurse providing care inserted a soft swab into the infant’s mouth for 3 min to collect saliva for cortisol assay. Timing of cortisol sampling allowed for the 15- to 20-min delay that occurs between adrenal release of cortisol and its presence in saliva. Timing improved the likelihood of achieving an accurate baseline assessment and capturing peak reactivity to the caregiving stressor. Study personnel completed review of the medical record at 2–4 weeks postnatal to acquire data on exposure to antenatal corticosteroids. We obtained data on GA, birth weight (BW), and neonatal morbidity from the record to control for these variables as necessary in the statistical analysis. We used information on neonatal morbidity to complete the Morbidity Assessment Index for Newborns (MAIN; see Measures section). In addition, mothers completed a sociodemographic questionnaire.
Measures Salivary cortisol. Saliva was collected using the infant swab from Salimetrics. Study protocol initially called for the swab to be left in the infant’s mouth for 1 min, but this time period produced inadequate amounts of saliva for analysis of many baseline or postcaregiving samples. As a result, we increased the time for saliva collection to 3 min. Samples were frozen and stored in the Pediatric Clinical Research Center until we sent them to Salimetrics for cortisol assay. Samples were assayed in duplicate to determine cortisol levels using a highly sensitive enzyme immunoassay (Salimetrics, State College, PA, www.salimetrics.com). The test used 25 μl of saliva per determination and has a lower limit of sensitivity of 0.007 μg/dl, standard curve range from 0.012 μg/dl to 3.0 μg/dl, an average intraassay coefficient of variation of 4.6%, and an average inter-assay coefficient of variation of 5.9%. Method accuracy determined by spike and recovery averaged 105.3% and linearity determined by serial dilution averaged 105.3%. Values from matched serum and saliva samples show a strong linear relationship, r (47) ¼ .91, p < .0001. HR. Infant HR was obtained from NICU monitors that continuously record physiologic indicators for each baby. An RA recorded the HR for 1-min epochs during both the baseline and the postcaregiving periods. We created two scores from these
data, namely, the mean of all minutes during the baseline period and the mean of all minutes during postcaregiving. HR monitors in the NICU (General Electric Solar 8000) are calibrated at the start of every shift. Nurses auscultate HR and compare it to the monitors to assure accuracy; monitors are adjusted as needed. Nurses also ensure accurate placement of electrodes before initiating the care. The accuracy of the neonatal monitors was based on the sample resolution and HR calculation algorithms, with stated specifications of +1% or +1 beats per minute [bpm], whichever is greater. The engineering team regularly tests the accuracy, functionality, and electrical safety of the electrocardiogram (ECG) monitors to ensure that they meet these manufacturer specifications. MAIN. The MAIN (Verma, Weir, Drummond, & Mitchell, 2005) measures the severity of an infant’s health problems. It contains 47 binary items representing 24 attributes of early neonatal morbidity such as complications from fetal drug exposure or intraventricular hemorrhage. Weights are assigned to items that were found through previous research to reflect overall morbidity based on predictive analyses with a sample of approximately 3,000 newborns. A sum score is computed by aggregating scale values of all items in the inventory. Scores of 2,000 moderate-to-severe morbidity. Verma, Weir, Drummond, and Mitchell (2005) report good validity and reliability of the measure. We scored items for each infant in our sample using data gathered from medical records. Antenatal corticosteroids. We determined exposure to antenatal corticosteroids by review of the electronic medical record of each infant. Data acquired from the record included the dose of the drug given, the number of courses received, and the week of gestation during which the infant was exposed. Mothers of each exposed infant had received two injections of betamethasone (12 mg each) during their third trimester; doses were given 24 hr apart. Sociodemographic questionnaire. We collected information on infant gender, race and ethnicity, family income, and use of government assistance for descriptive purposes.
Data Analysis Descriptive statistics were used to examine sample characteristics. A repeated-measures (mixed design) analysis of covariance was employed to determine the effect of antenatal corticosteroids on both cortisol and HR reactivity to caregiving. Covariates controlled for in the analysis were infant BW and neonatal morbidity. Data were analyzed with SPSS version 21.
Downloaded from brn.sagepub.com at UNIV NEBRASKA LIBRARIES on September 29, 2015
4
Biological Research for Nursing
Table 1. Infant Characteristics by Corticosteroid Group. Corticosteroids n ¼ 42
No corticosteroids n ¼ 15
Female sex (%) 45 Gestational age (weeks) 32.6 (1.79) Birth weight (g) 1,765 (515.72) 0.24 (0.18) Baseline cortisol (mg/dl)a Postcaregiving cortisol (mg/dl)a 0.25 (0.18) Baseline heart rate (bpm) 156 (14) Postcaregiving heart rate (bpm) 155 (13)
68 34.7 (1.58) 2,261 (413.80) 0.42 (0.06) 0.56 (0.15) 138 (15) 145 (18)
Characteristic
Note. bpm ¼ beats per minute. Data are provided as mean (SD) unless otherwise noted. a Cortisol measures were available for only n ¼ 13 infants exposed to antenatal corticosteroids and n ¼ 5 infants who were not exposed.
Results Characteristics of the Sample The sample consisted of 57 infants (29 females and 28 males) receiving care in the NICU. The mean GA of the infants was 33.27 weeks, ranging from 29 weeks to 36 weeks. On average, their BW was 1,916 g, with a range from 785 g to 2,825 g. For severity of their illness, infants’ scores ranged from 0 to 2,633, with a mean of 614. Of the 57 infants, 74% (n ¼ 42) had received antenatal corticosteroids. Three of these infants had a second course of corticosteroids but they did not differ from other infants who received only 1 course of corticosteroids on either baseline or recovery means of cortisol and HR. The mean fetal age at which corticosteroids were given was 30 (SD ¼ 2.8) weeks, ranging from 23 weeks to 35 weeks. We provide infant characteristics by corticosteroid group in Table 1. A slightly larger percentage of the infants exposed to corticosteroids was male. The mean GA of the exposed group was 32.6 weeks (range 29.2–35.6) in contrast to 34.7 weeks (range 29.5–36.0) in the nonexposed group (t ¼ 3.73, df ¼ 56, p ¼ .001). Average BW of the exposed group was 1,765 g (range 785–2,825) in contrast to 2,261 g (range 1,360–2,805) in the nonexposed group (t ¼ 3.35, df ¼ 56, p ¼ .001). Exposed infants (M ¼ 707) also had greater severity of illness at birth than nonexposed infants (M ¼ 437, t ¼ 2.63, df ¼ 56, p ¼ .01). Self-reported family heritage was Caucasian (75%), African (12%), Asian (7%), and other (6%). A total of 17% of the families reported Hispanic ethnicity. Regarding income, 21% of the families earned less than US$15,000 per year, 14% earned US$15,000–US$50,000 per year, 8% earned US$51,000– US$100,000, 8% earned US$101,000–US$149,000, and 11% earned US$150,000 or more. A total of 38% of the families received some form of government assistance.
Testing of the Research Aim Group differences in GA, BW, and severity of neonatal illness necessitated that we control for these in our analysis. However, a very high correlation between GA and BW (r ¼ .78, p < .000)
Table 2. Effects of Corticosteroid Exposure During Gestation on Cortisol Reactivity to Caregiving, Controlling for Birth Weight, and Neonatal Morbidity. Variable
df
MS
F
Partial Z2
Birth weight Neonatal morbidity Corticosteroid exposure
1 1 1
.25 .00 .28
10.79* 0.16 12.37**
.44 .01 .47
Note. n ¼ 18; MS ¼ mean square. *p ¼ .005. **p ¼ .003.
suggested that they were measuring very similar information about the infant. In order to avoid Type II error, we included only one of these variables in the analysis. We selected BW because of its stronger bivariate correlation with our baseline measures of HR and cortisol response. Severity of illness was only modestly related to both GA (r ¼ .40, p ¼ .002) and BW (r ¼ .34, p ¼ .008). Cortisol response. Because of the data collection problems described earlier, cortisol data were available from only 18 of the 57 infants. Of these, 13 had been exposed to corticosteroids and 5 had not. Means and standard deviations for cortisol values of the two groups are shown in Table 1. Results of the analysis for effect of corticosteroids on cortisol response are presented in Table 2. BW was significantly and positively related to cortisol values, but neonatal morbidity was not. After controlling for any effect these variables might have, infants who were exposed to corticosteroids during gestation differed significantly in their cortisol response from infants who were not exposed, F(1,14) ¼ 12.37, p ¼ .003. As shown in Table 1, baseline cortisol levels were substantially lower for infants who were exposed to corticosteroids (M mg/dl ¼ .24) than for infants who were not exposed (M mg/dl ¼ .42). In addition, exposed infants showed no change in response to the caregiving in their cortisol level, F(1,15) ¼ .54, not significant (NS), while infants who were not exposed to steroids showed a significant increase in cortisol level as a result of the caregiving, F(1,4) ¼ 67.60, p
Effects of Antenatal Corticosteroids on Cortisol and Heart Rate Reactivity of Preterm Infants
Biological Research for Nursing 1-8 ª The Author(s) 2015 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/1099800414564860 brn.sagepub.com
Sandra J. Weiss, PhD, DNSc, RN, FAAN1, and Sandra Niemann, PhD, RN1
Abstract Administration of glucocorticoids (GCs) during pregnancy is an established practice for reducing morbidity and mortality of fetuses at risk of preterm delivery. However, preliminary research indicates that exposure to exogenous GCs in utero may be associated with suppressed hypothalamic–pituitary–adrenal axis activity. The aim of this study was to determine whether preterm neonates who are exposed to antenatal corticosteroids show evidence of a suppressed stress-response system during their first few weeks of life, in contrast to infants who are not exposed. The sample (51% female) included 57 neonates, with 74% exposed to steroids. Mean gestational ages (GAs) were 32.6 weeks for exposed and 34.7 weeks for nonexposed infants. Although neonates in the two groups differed in gender, birth weight, and morbidity, these factors were controlled for in data analyses. Infants’ salivary cortisol and heart rate (HR) were measured before and after they received a standardized caregiving ‘‘stressor’’ while in the newborn intensive care unit. Infants exposed to GCs in utero had lower basal cortisol levels and higher HRs than their nonexposed peers. In contrast to infants who received no GCs, they also exhibited minimal HR or cortisol reactivity to the stressor. Findings suggest that preterm infants who were exposed to antenatal corticosteroids experience a suppressed response to stress. As preterm children develop, this dysregulation has numerous implications for later development of stress-related cardiovascular and mental health problems. Further research is needed to determine whether these postnatal effects of antenatal corticosteroids persist over time. Keywords corticosteroids, HPA axis, infant stress response, prematurity
Preterm birth is a major health issue for families and for the health care system. Approximately 12% of the U.S. births are preterm (Martin et al., 2012). Prematurity represents a risk factor for biobehavioral and neurodevelopmental disorders that lead to multiple health problems in childhood and adulthood (Gluckman & Hansen, 2004; Gonc¸alves, Chaiworapongsa, & Romero, 2002; Thompson, Syddall, Rodin, Osmond, & Barker, 2001). Infant morbidity associated with prematurity clearly contributes to these problems; however, well-meaning treatments may inadvertently influence these outcomes as well. For example, it has become the standard of care in obstetric practice to give antenatal corticosteroids to women at risk of preterm birth (Gilstrap et al., 1995; Roberts & Dalziel, 2006). Although the short-term benefits of this treatment are widely recognized, recent findings raise concerns regarding its effects on development of the fetal hypothalamic–pituitary–adrenal (HPA) axis and autonomic nervous system (ANS), interrelated structures that play key roles in determining resilience or vulnerability to stress. Administration of betamethasone or dexamethasone, the two most commonly used antenatal corticosteroids, is designed
to mimic the rise in cortisol that occurs shortly before birth (Rog-Zielinska, Richardson, Denvir, & Chapman, 2014). As synthetic corticosteroids, however, they have somewhat different properties than endogenously produced glucocorticoids (GCs). Most importantly, synthetic corticosteroids are not deactivated by the placental enzyme 11B-HSD2 (Alexander et al., 2012), thus putting the fetal brain at risk for substantial exposure to the drugs. Recent research using rigorously controlled animal models has raised concerns about the impact of exposure to antenatal corticosteroids on development of the HPA axis. Exposure to antenatal steroids has been associated with attenuated fetal GC secretion, blunted function of the HPA
1
Department of Community Health Systems, School of Nursing, University of California, San Francisco, CA, USA Corresponding Author: Sandra Weiss, PhD, DNSc, RN, Department of Community Health Systems, School of Nursing, University of California, Box 0608, 2 Koret Way, San Francisco, CA 94143, USA. Email: [email protected]
Downloaded from brn.sagepub.com at UNIV NEBRASKA LIBRARIES on September 29, 2015
2
Biological Research for Nursing
axis, and permanent alteration of HPA-axis reactivity later in life (Burlet et al., 2005; Kapoor, Petropoulos, & Mathews, 2008; Shoener, Baig, & Page, 2006). Although studies are few, research indicates that human infants are also affected by synthetic GCs. For term infants, stress-induced responses measured at 24 hr postnatal (Davis, Waffarn, & Sandman, 2011) indicated larger cortisol reactivity in infants exposed to GCs than in infants not exposed. In four studies of preterm infants exposed to GCs, investigators examined infants’ stress responses at 3–6 days postnatal (Ashwood et al., 2006; Davis et al., 2004, 2006; Schaffer, Luzi, Burkhardt, Rauh, & Beinder, 2009), with Davis et al.’s (2006) study also assessing responses during the infants’ second week postnatal. Results for both the Davis et al. and Schaffer et al. studies indicated a suppressed cortisol response in the preterm infants exposed to GCs, in contrast to preterm infants who were not exposed. Ashwood et al. found that 38% of preterm infants receiving a single course of GCs had a suppressed cortisol response, while 68% of infants receiving a repeat course showed suppression. The only other study of which we are aware found that antenatal GCs predicted a suppressed cortisol response at 4 months postnatal in response to immunization (Glover, Miles, Matta, Modi, & Stevenson, 2005). These consistent findings of suppression among GC-exposed preterm infants could stem from enhanced negative feedback on the HPA axis as a result of fetal GC exposure and/or increased tissue sensitivity to GCs (Harris & Seckl, 2011). The HPA axis acts in close concert with the ANS in response to stress (Laurent, Powers, & Granger, 2013). Both systems are regulated by neurons in the paraventricular nucleus of the hypothalamus and have bidirectional effects (Ulrich-Lai & Herman, 2009). The sympathetic nervous system directly innervates the adrenal cortex, which controls synthesis and release of GC hormones. Circulating GCs, in turn, can also potentiate sympathetic nervous system changes, including increased heart rate (HR). Thus, administration of antenatal corticosteroids may also affect key autonomic indicators such as HR. Only a few studies have examined these immediate or longterm effects, however. Davis et al. (2004) reported that HR increased significantly (p ¼ .0001) in response to a heel-stick stressor in steroid-exposed infants compared to controls at 3–6 days postnatal, although the overall HR did not differ between groups. These same infants had a suppressed cortisol response to the heel-stick stressor, suggesting that antenatal steroids may exert opposite early effects on the neuroendocrine and autonomic systems. Other researchers who investigated longer term effects of fetal exposure to steroids found no significant differences in HR between treatment groups for children 7–10 years of age (de Vries et al., 2008) or 30-yearold adults (Dalziel et al., 2005). However, neither of these studies examined HR reactivity to stress. This pilot study was conducted to determine the feasibility of data collection procedures and initial effect sizes for estimating sample size needs of a larger study. It also adds to our knowledge about the effects of antenatal corticosteroids on the HPA axis and ANS in the immediate postnatal period. The aim
of this research was to determine whether preterm infants who are exposed to corticosteroids during gestation show a suppressed stress-response system during the first few weeks of life, as evidenced by dampened cortisol response and HR reactivity. Our hypothesis was that infants exposed to corticosteroids during gestation will have less cortisol reactivity and less HR reactivity in response to a stressor than infants who are not exposed to corticosteroids. Although our hypothesis regarding cortisol is congruent with most previous research, our hypothesis related to HR is not informed by the one previous study showing increased HR for steroid-exposed infants in response to a stressor. Instead, our HR hypothesis was based on previous research regarding the strong, positive relationship between cortisol and HR reactivity in preterm infants (Haley, Grunau, Weinberg, Keidar, & Oberlander, 2010) and the established interconnectedness of the two regulatory systems (e.g., Janig, 2006; Stalder, Evans, Hucklebridge, & Clow, 2011). Thus, we expected to see a parallel coupling of physiologic responses and a general pattern of suppression across both autonomic and neuroendocrine systems.
Method Procedures Infants were eligible to participate if they were born between 28 and 37 weeks’ gestation and had no known brain malformation, genetic anomaly, or risk of dying. In the obstetrics service at our institution, between 70% and 80% of women who deliver preterm infants are given antenatal corticosteroids by their clinicians. As a result, we did not expect the sample to be evenly distributed between groups of infants who were exposed to corticosteroids and those who were not. Research nurses in the Pediatric Clinical Research Center identified eligible infants in the newborn intensive care unit (NICU) and provided a flyer about the study to their mothers. Mothers who indicated an interest in participating were approached by the research team to obtain written consent. This study was approved by the Institutional Review Board of the university. We assessed each infant’s stress response (cortisol and HR reactivity) twice in the NICU 2–4 weeks after birth. The stressor included 10 min of care provided by a neonatal nurse. Caregiving included changing the infant’s diaper, performing peri care, taking the temperature, checking the feeding tube, and measuring abdominal girth. The protocol for the caregiving stressor was based upon standard care provided to infants in the NICU and developed in consultation with the NICU nurses. This process assured that all nurses performing the care would be familiar with the protocol and comfortable with its safety for the infants. We held a meeting with the nurses at the beginning of the study to review the protocol and answer any questions about its implementation. To enhance fidelity, we gave the nurse providing the care a small card to review at the start of each data collection session that highlighted the components of the protocol and the order in which they should occur. The card remained near the crib or incubator during the entire
Downloaded from brn.sagepub.com at UNIV NEBRASKA LIBRARIES on September 29, 2015
Weiss and Niemann
3
procedure for reference, if needed. All caregiving sessions were scheduled between 2 and 4 p.m. in the afternoon, one of the quietest times in the NICU when procedures, interventions, and visitors are minimal. The infant rested without any contact for 15 min prior to data collection. If the infant was not in a quiet state, baseline time was extended to assure a 15-min period at rest. A research assistant (RA) then collected baseline HR data for 5 min before caregiving began. After caregiving, the infant was swaddled and allowed to rest quietly for a 5-min postcaregiving period. The RA again collected HR data during this time. At the end of the baseline period and again after the postcaregiving period ended, the nurse providing care inserted a soft swab into the infant’s mouth for 3 min to collect saliva for cortisol assay. Timing of cortisol sampling allowed for the 15- to 20-min delay that occurs between adrenal release of cortisol and its presence in saliva. Timing improved the likelihood of achieving an accurate baseline assessment and capturing peak reactivity to the caregiving stressor. Study personnel completed review of the medical record at 2–4 weeks postnatal to acquire data on exposure to antenatal corticosteroids. We obtained data on GA, birth weight (BW), and neonatal morbidity from the record to control for these variables as necessary in the statistical analysis. We used information on neonatal morbidity to complete the Morbidity Assessment Index for Newborns (MAIN; see Measures section). In addition, mothers completed a sociodemographic questionnaire.
Measures Salivary cortisol. Saliva was collected using the infant swab from Salimetrics. Study protocol initially called for the swab to be left in the infant’s mouth for 1 min, but this time period produced inadequate amounts of saliva for analysis of many baseline or postcaregiving samples. As a result, we increased the time for saliva collection to 3 min. Samples were frozen and stored in the Pediatric Clinical Research Center until we sent them to Salimetrics for cortisol assay. Samples were assayed in duplicate to determine cortisol levels using a highly sensitive enzyme immunoassay (Salimetrics, State College, PA, www.salimetrics.com). The test used 25 μl of saliva per determination and has a lower limit of sensitivity of 0.007 μg/dl, standard curve range from 0.012 μg/dl to 3.0 μg/dl, an average intraassay coefficient of variation of 4.6%, and an average inter-assay coefficient of variation of 5.9%. Method accuracy determined by spike and recovery averaged 105.3% and linearity determined by serial dilution averaged 105.3%. Values from matched serum and saliva samples show a strong linear relationship, r (47) ¼ .91, p < .0001. HR. Infant HR was obtained from NICU monitors that continuously record physiologic indicators for each baby. An RA recorded the HR for 1-min epochs during both the baseline and the postcaregiving periods. We created two scores from these
data, namely, the mean of all minutes during the baseline period and the mean of all minutes during postcaregiving. HR monitors in the NICU (General Electric Solar 8000) are calibrated at the start of every shift. Nurses auscultate HR and compare it to the monitors to assure accuracy; monitors are adjusted as needed. Nurses also ensure accurate placement of electrodes before initiating the care. The accuracy of the neonatal monitors was based on the sample resolution and HR calculation algorithms, with stated specifications of +1% or +1 beats per minute [bpm], whichever is greater. The engineering team regularly tests the accuracy, functionality, and electrical safety of the electrocardiogram (ECG) monitors to ensure that they meet these manufacturer specifications. MAIN. The MAIN (Verma, Weir, Drummond, & Mitchell, 2005) measures the severity of an infant’s health problems. It contains 47 binary items representing 24 attributes of early neonatal morbidity such as complications from fetal drug exposure or intraventricular hemorrhage. Weights are assigned to items that were found through previous research to reflect overall morbidity based on predictive analyses with a sample of approximately 3,000 newborns. A sum score is computed by aggregating scale values of all items in the inventory. Scores of 2,000 moderate-to-severe morbidity. Verma, Weir, Drummond, and Mitchell (2005) report good validity and reliability of the measure. We scored items for each infant in our sample using data gathered from medical records. Antenatal corticosteroids. We determined exposure to antenatal corticosteroids by review of the electronic medical record of each infant. Data acquired from the record included the dose of the drug given, the number of courses received, and the week of gestation during which the infant was exposed. Mothers of each exposed infant had received two injections of betamethasone (12 mg each) during their third trimester; doses were given 24 hr apart. Sociodemographic questionnaire. We collected information on infant gender, race and ethnicity, family income, and use of government assistance for descriptive purposes.
Data Analysis Descriptive statistics were used to examine sample characteristics. A repeated-measures (mixed design) analysis of covariance was employed to determine the effect of antenatal corticosteroids on both cortisol and HR reactivity to caregiving. Covariates controlled for in the analysis were infant BW and neonatal morbidity. Data were analyzed with SPSS version 21.
Downloaded from brn.sagepub.com at UNIV NEBRASKA LIBRARIES on September 29, 2015
4
Biological Research for Nursing
Table 1. Infant Characteristics by Corticosteroid Group. Corticosteroids n ¼ 42
No corticosteroids n ¼ 15
Female sex (%) 45 Gestational age (weeks) 32.6 (1.79) Birth weight (g) 1,765 (515.72) 0.24 (0.18) Baseline cortisol (mg/dl)a Postcaregiving cortisol (mg/dl)a 0.25 (0.18) Baseline heart rate (bpm) 156 (14) Postcaregiving heart rate (bpm) 155 (13)
68 34.7 (1.58) 2,261 (413.80) 0.42 (0.06) 0.56 (0.15) 138 (15) 145 (18)
Characteristic
Note. bpm ¼ beats per minute. Data are provided as mean (SD) unless otherwise noted. a Cortisol measures were available for only n ¼ 13 infants exposed to antenatal corticosteroids and n ¼ 5 infants who were not exposed.
Results Characteristics of the Sample The sample consisted of 57 infants (29 females and 28 males) receiving care in the NICU. The mean GA of the infants was 33.27 weeks, ranging from 29 weeks to 36 weeks. On average, their BW was 1,916 g, with a range from 785 g to 2,825 g. For severity of their illness, infants’ scores ranged from 0 to 2,633, with a mean of 614. Of the 57 infants, 74% (n ¼ 42) had received antenatal corticosteroids. Three of these infants had a second course of corticosteroids but they did not differ from other infants who received only 1 course of corticosteroids on either baseline or recovery means of cortisol and HR. The mean fetal age at which corticosteroids were given was 30 (SD ¼ 2.8) weeks, ranging from 23 weeks to 35 weeks. We provide infant characteristics by corticosteroid group in Table 1. A slightly larger percentage of the infants exposed to corticosteroids was male. The mean GA of the exposed group was 32.6 weeks (range 29.2–35.6) in contrast to 34.7 weeks (range 29.5–36.0) in the nonexposed group (t ¼ 3.73, df ¼ 56, p ¼ .001). Average BW of the exposed group was 1,765 g (range 785–2,825) in contrast to 2,261 g (range 1,360–2,805) in the nonexposed group (t ¼ 3.35, df ¼ 56, p ¼ .001). Exposed infants (M ¼ 707) also had greater severity of illness at birth than nonexposed infants (M ¼ 437, t ¼ 2.63, df ¼ 56, p ¼ .01). Self-reported family heritage was Caucasian (75%), African (12%), Asian (7%), and other (6%). A total of 17% of the families reported Hispanic ethnicity. Regarding income, 21% of the families earned less than US$15,000 per year, 14% earned US$15,000–US$50,000 per year, 8% earned US$51,000– US$100,000, 8% earned US$101,000–US$149,000, and 11% earned US$150,000 or more. A total of 38% of the families received some form of government assistance.
Testing of the Research Aim Group differences in GA, BW, and severity of neonatal illness necessitated that we control for these in our analysis. However, a very high correlation between GA and BW (r ¼ .78, p < .000)
Table 2. Effects of Corticosteroid Exposure During Gestation on Cortisol Reactivity to Caregiving, Controlling for Birth Weight, and Neonatal Morbidity. Variable
df
MS
F
Partial Z2
Birth weight Neonatal morbidity Corticosteroid exposure
1 1 1
.25 .00 .28
10.79* 0.16 12.37**
.44 .01 .47
Note. n ¼ 18; MS ¼ mean square. *p ¼ .005. **p ¼ .003.
suggested that they were measuring very similar information about the infant. In order to avoid Type II error, we included only one of these variables in the analysis. We selected BW because of its stronger bivariate correlation with our baseline measures of HR and cortisol response. Severity of illness was only modestly related to both GA (r ¼ .40, p ¼ .002) and BW (r ¼ .34, p ¼ .008). Cortisol response. Because of the data collection problems described earlier, cortisol data were available from only 18 of the 57 infants. Of these, 13 had been exposed to corticosteroids and 5 had not. Means and standard deviations for cortisol values of the two groups are shown in Table 1. Results of the analysis for effect of corticosteroids on cortisol response are presented in Table 2. BW was significantly and positively related to cortisol values, but neonatal morbidity was not. After controlling for any effect these variables might have, infants who were exposed to corticosteroids during gestation differed significantly in their cortisol response from infants who were not exposed, F(1,14) ¼ 12.37, p ¼ .003. As shown in Table 1, baseline cortisol levels were substantially lower for infants who were exposed to corticosteroids (M mg/dl ¼ .24) than for infants who were not exposed (M mg/dl ¼ .42). In addition, exposed infants showed no change in response to the caregiving in their cortisol level, F(1,15) ¼ .54, not significant (NS), while infants who were not exposed to steroids showed a significant increase in cortisol level as a result of the caregiving, F(1,4) ¼ 67.60, p
