Psychiatric Quarterly

, Volume 84, Issue 4, pp 475–484

The Association Between Salivary Hormone Levels and Children’s Inpatient Aggression: A Pilot Study

Authors

    • Division of Child and Adolescent PsychiatryCincinnati Children’s Hospital Medical Center
  • Douglas Mossman
    • Department of Psychiatry and Behavioral NeuroscienceUniversity of Cincinnati College of Medicine
  • Kacey Appel
    • Cincinnati Children’s Hospital Medical Center
  • Thomas J. Blom
    • Department of Psychiatry and Behavioral NeuroscienceUniversity of Cincinnati College of Medicine
  • Jeffrey R. Strawn
    • Department of Psychiatry and Behavioral NeuroscienceUniversity of Cincinnati College of Medicine
  • Nosa N. Ekhator
    • Department of Psychiatry and Behavioral NeuroscienceUniversity of Cincinnati College of Medicine
  • Bianca Patel
    • Case-Western Reserve University
  • Melissa P. DelBello
    • Department of Psychiatry and Behavioral NeuroscienceUniversity of Cincinnati College of Medicine
  • Michael Sorter
    • Division of Child and Adolescent PsychiatryCincinnati Children’s Hospital Medical Center
  • David Klein
    • Division of EndocrinologyCincinnati Children’s Hospital Medical Center
  • Thomas D. GeraciotiJr.
    • Department of Psychiatry and Behavioral NeuroscienceUniversity of Cincinnati College of Medicine
Original Paper

DOI: 10.1007/s11126-013-9260-8

Cite this article as:
Barzman, D.H., Mossman, D., Appel, K. et al. Psychiatr Q (2013) 84: 475. doi:10.1007/s11126-013-9260-8

Abstract

Aggression is a common management problem for child psychiatry hospital units. We describe an exploratory study with the primary objective of establishing the feasibility of linking salivary concentrations of three hormones (testosterone, dehydroepiandrosterone [DHEA], and cortisol) with aggression. Between May 2011 and November 2011, we recruited 17 psychiatrically hospitalized boys (age 7–9 years). We administered the Brief Rating of Aggression by Children and Adolescents (BRACHA) and Predatory-Affective Aggression Scale (PAAS) upon admission. Saliva samples were collected from the participants during a 24-h period shortly after admission: immediately upon awakening, 30 min later, and again between 3:45 and 7:45 P.M. Nursing staff recorded Overt Aggression Scale ratings twice a day during hospitalization to quantify aggressive behavior. The salivary cortisol concentrations obtained from aggressive boys 30 min after awakening trended higher than levels from the non-aggressive boys (p = 0.06), were correlated with the number of aggressive incidents (p = 0.04), and trended toward correlation with BRACHA scores (p = 0.06). The aggressive boys also showed greater morning-to-evening declines in cortisol levels (p = 0.05). Awakening levels of DHEA and testosterone were correlated with the severity of the nearest aggressive incident (p < 0.05 for both). The BRACHA scores of the aggressive boys were significantly higher than scores of the non-aggressive boys (p < 0.001). Our data demonstrate the feasibility of collecting saliva from children on an inpatient psychiatric unit, affirm the utility of the BRACHA in predicting aggressive behavior, and suggest links between salivary hormones and aggression by children who undergo psychiatric hospitalization.

Keywords

AggressionBRACHAHormonesChildPsychiatric hospitalization

Introduction

Aggressive behavior is a common and important management problem for child psychiatry hospital units. To date, clinicians have developed few tools to aid in identifying which patients will behave aggressively. The Brief Rating of Aggression by Children and Adolescents (BRACHA) can help categorize children and adolescents into distinct higher and lower aggression risk groups during hospitalization [1], and the Predatory-Affective Aggression Scale (PAAS) [2] can potentially separate psychiatrically-hospitalized children into predatory and affective aggression clusters. Yet an objective, non-invasive biological test would be a useful supplement to clinician-administered rating scales in predicting aggressive behavior and in understanding distinctive features or clinical profiles related to aggressive behavior in hospitals.

Previous reports have explored hormone levels and pediatric aggression, but to our knowledge, no studies have prospectively evaluated differences in hormone levels in groups at high and low risk for aggression or the usefulness salivary hormones in predicting aggression on pediatric inpatient units [3]. Hormone levels obtained from saliva samples have advantages over levels obtained from other biological fluids (e.g., cerebrospinal fluid, blood, or urine), particularly when evaluations of children are involved. Collection of saliva is noninvasive, the stress response to sample collection is minimal, and the level of cooperation required is modest. In addition, measurements from saliva are accurate and reproducible in determining the peripheral concentrations of cortisol, dehydroepiandrosterone (DHEA), and testosterone [4]. Finally, saliva can be easily collected at closely spaced intervals.

We conducted a pilot study to establish the feasibility of studying salivary hormone levels as they relate to aggression in 7–9 year olds. We selected this narrow age range to avoid the developmental impulsiveness of preschoolers and the hormonal changes associated with puberty. In this pilot study, we used BRACHA scores to select children at low and high risk for aggression to increase the likelihood of selecting patients from both aggressive and nonaggressive subgroups. We also examined the relationship between participant’s BRACHA scores (risk level) and salivary hormone levels as possible clues to the biology of pediatric aggression.

Methods

Participants

This study received approval from the Institutional Review Board at Cincinnati Children's Hospital Medical Center. All participants and their legal guardians demonstrated understanding of the study and provided informed assent and consent, respectively. Inclusion criteria restricted participants to boys aged 7–9 years who had undergone screening with the BRACHA prior to their admission to one of the psychiatric inpatient units. The BRACHA is a 14-item instrument scored by emergency room social workers to help assess risk for pediatric aggression during an upcoming psychiatric hospitalization [1]. Recent findings suggest that the total BRACHA score produces highly reliable ranks of the risk of aggression by children and adolescents who are being psychiatrically admitted [5]. Exclusion criteria included presence of viral or bacterial infection requiring antibiotics within 2 weeks of admission, recent surgery within 8 weeks of admission, bleeding gums within 8 weeks of admission, current detainment in juvenile detention, and steroid use. None of the participants had undergone incarceration in a juvenile detention facility. All participants and guardians who were asked to participate agreed to do so, and no participant dropped out of the study. No participant refused to provide saliva or participate and no data was collected on excluded participants.

Between May 2011 and November 2011, we recruited 17 psychiatrically hospitalized boys (age 7–9 years). We collected data from participants who met the inclusion criteria from consecutive admissions. We categorized the 17 participants into two groups, high aggression risk and low aggression risk, based on their initial BRACHA score, which was obtained per usual hospital practice by trained psychiatric social workers within the pediatric emergency department at CCHMC. The high-risk group was defined as having a BRACHA score ≥8; the low-risk group had scores of ≤6.5. We excluded potential participants with BRACHA scores of 7 or 7.5, which previous research showed was indicative of intermediate aggression risk. The high-risk group consisted of 9 boys and the low-risk group consisted of 8 boys. Psychiatric diagnoses for all participants were established, per usual clinical practice, by each child’s treating psychiatrist based on all available clinical information at the conclusion of hospitalization (see Table 1).We did not determine the primary diagnosis. Eleven (65 %) of the 17 participants were diagnosed with a mood disorder and 9 (53 %) were diagnosed with ADHD.
Table 1

Diagnoses for participants

Participant

Diagnosis 1

Diagnosis 2

Diagnosis 3

Diagnosis 4

1

DBD

BIF

MD

None

2

MD

ODD

IED

None

3

MD

OSD

ADHD

None

4

MPDR

L

PTSD

ADHD, CT

5

MD

ADHD, CT

None

None

6

MD

PTSD

ADHD, CT

None

7

DBD

L

None

None

8

IED

None

None

None

9

IED

Depression

ADHD, CT

None

10

MD

PTSD

ODD

ADHD, CT

11

IED

ODD

MD

None

12

IED

None

None

None

13

MD

ADHD, CT

None

None

14

DBD

ADHD, CT

None

None

15

MD

L

PTSD

None

16

MDSR, no P

ADHD, CT

None

None

17

IED

L

PCP

None

Legend

Description

DSM-IV-TR code

ADHD, CT

Attention-deficit/hyperactivity disorder, combined type

314.01

BIF

Borderline intellectual functioning

V62.89

DBD

Disruptive behavior disorder

312.9

IED

Intermittent explosive disorder

312.34

L

Language impairment

784.59

MD

Mood disorder

296.90

MDSR

Major depression recurrent, severe, without psychotic features

296.33

MPDR

Major psychotic depression, recurrent

296.34

ODD

Oppositional defiant disorder

313.81

OSD

Other speech disturbance

784.59

PCP

Parent–child problem

V61.20

PTSD

Posttraumatic stress disorder

309.81

Ratings

Each participant underwent evaluation with the PAAS in addition to the BRACHA. Overt Aggression Scale (OAS) ratings were obtained for each patient twice a day after the daytime and evening shifts. The OAS is a rating scale for verbal and physical aggression toward self, other persons, and objects [6]. If an incident of aggression occurs, the rater assigns a score on a four-point severity scale anchored with behavioral descriptors. Several studies have used the OAS as a reliable and appropriate measure for quantifying aggression on a child psychiatric unit [79]. We also obtained each participant’s height, weight, and medication record.

The primary outcome measure was absence or presence of an aggressive incident during inpatient admission, as determined by the OAS. We categorized as “aggressive” those patients who engaged in behavior that produced a score of 1 or more on any subscale of the OAS (aggression towards other people, aggression towards objects, self-aggression, and verbal aggression) during the hospitalization. Severity of nearest aggressive incident was based on the highest rating (1–4) of the four OAS subtypes nearest in time to saliva sampling. The “not aggressive” group had scores of 0 on the OAS throughout the hospitalization.

Saliva Sampling

We collected three saliva samples over a 24-h period on one of the initial three hospital days. The first sample was obtained immediately upon awakening (before eating or teeth brushing; n = 17); the second was obtained 28–30 min later (n = 15); and the final sample was obtained between 3:45 pm and 7:45 pm (n = 15), depending on when the participant last consumed food. Collection of saliva samples in polypropylene tubes followed 60 s of oral rinsing. Participants were instructed to drool approximately 5 ml into the tube, and the saliva-containing tubes were frozen within 5 min of collection. Two second morning samples and two late afternoon samples were not obtained due to participants’ having eaten during the previous hour.

Salivary Hormone Assays

A commercially available competitive enzyme-linked immunoassay (Salimetrics LLC, State College, PA) was used to quantify salivary cortisol, DHEA, and testosterone. All reagents, including microtitre plates coated with rabbit antibodies to cortisol, DHEA, and testosterone, were obtained from Salimetrics LLC. On the day of assay, all saliva samples, which had been previously frozen at −80 °C, were gradually thawed at room temperature for fewer than 5 min. After thawing, each tube was thoroughly vortexed for 50 s. Next, 25 µl (cortisol, testosterone) or 50 µl (DHEA) of saliva was transferred into microtitre plate wells coated with rabbit antibody to each analyte. Two-hundred μl of enzyme conjugate solution (horseradish peroxidase conjugated to antirabbit IgG goat antibody of the respective analyte) were pipetted into microwells, with continuous mixing for 5 min at 500 rpm. Plates were incubated for 55 min (cortisol and testosterone) or 3 h (for DHEA) at room temperature. Following the incubation, the unbound components were washed with buffer solution. After washing, 200 μl of colorimetric 3,3′,5,5′-tetramethylbenzidine (TMB) were added to each well and mixed for 5 min at room temperature. Thereafter, each plate was incubated at room temperature in darkness for 25 min. To stop the reaction, 50 μl of 2 M H2SO4 were added to each well. Optical density (OD) was read using a Tecan plate reader (Switzerland) at a wavelength of 450 nm. The intra- and inter-assay coefficients of variation for all analytes were under 8 %.

Data Analysis

We compared demographic and clinical characteristics in the aggressive group and non-aggressive group (independent variable) using Fisher’s exact tests for binary variables and Wilcoxon’s rank sum exact tests for continuous variables. We analyzed the salivary hormone concentrations over time with random intercepts logistic regression to account for the within-participant correlation. The dependent variable was high or low level for each specific hormone, as defined as above or below the overall median value. Predictors in these models were aggression group, time, and their interaction. Spearman correlation coefficients were calculated at each time point for the hormonal data versus aggression variables. All analyses were performed with SAS 9.2.

Results

Demographic and Clinical Characteristics

The aggressive and non-aggressive groups were demographically similar (Table 2). The BRACHA scores of the boys who exhibited aggression had almost no overlap with and were significantly higher than the BRACHA scores of the non-aggressive boys (p < 0.001). Five of the 10 aggressive boys were receiving α2-adrenergic agonists (clonidine and guanfacine), but none of the non-aggressive boys were (p < 0.05). However, the aggressive and non-aggression boys did not differ regarding other medications. The type of aggression, as measured by the PAAS, did not significantly differ in the two groups (p = 0.36).
Table 2

Demographic, clinical characteristics, and hormone levels by aggression group

 

No aggression

(n = 7)

Any aggression

(n = 10)

p*

Age (years)

8.4 (0.7)

8.0 (0.8)

0.28

Race (# Caucasian)

6 (86 %)

5 (50 %)

0.30

Height (cm)

132.2 (6.6)

132.0 (6.9)

0.83

Weight (kg)

32.8 (15.9)

32.8 (10.1)

0.74

BMI (kg/m2)

18.2 (6.8)

18.6 (4.6)

0.74

Living arrangement (# living w/one adult)

4 (57 %)

7 (70 %)

0.64

Active medications

 Atypical anti-psychotic

5 (71 %)

8 (80 %)

0.99

 Bupropion

2 (29 %)

3 (30 %)

0.99

 Clonidine/guanfacine

0

5 (50 %)

0.04

 Stimulant (for ADHD)

3 (43 %)

4 (40 %)

0.99

PAAS Score

-0.9 (2.1)

0.1 (2.1)

0.36

BRACHA Score

5.3 (1.2)

9.2 (1.7)

<0.001

Length of Stay (days)

7.7 (1.4)

9.3 (3.4)

0.45

Mean (SD) or # (%) shown

*Fisher’s exact test (binary variables) or Wilcoxon rank sum exact test (continuous variables)

Salivary Cortisol

The boys who exhibited aggression had second morning cortisol levels that trended higher than levels from the non-aggressive boys (p = 0.06) and the aggressive boys had greater morning-to-evening declines in cortisol levels (p = 0.05). As Table 3 shows, the boys’ BRACHA scores and second morning cortisol levels showed a trend towards a significant correlation (p = 0.06), and the number (but not the intensity) of aggressive incidents was significantly correlated with the second morning cortisol level (p = 0.04).
Table 3

Correlations between BRACHA, PAAS, aggression, and salivary hormones

 

Cortisol

DHEA

Testosterone

Time = AM1

 BRACHA score

−0.05 (p = 0.86)

−0.01 (p = 0.98)

0.40 (p = 0.11)

 PAAS score

−0.20 (p = 0.43)

−0.07 (p = 0.78)

−0.08 (p = 0.77)

 Number of aggressive incidents

−0.10 (p = 0.69)

−0.14 (p = 0.58)

0.24 (p = 0.36)

 Severity of nearest aggressive incident (n = 10)

0.33 (p = 0.35)

0.64 (p = 0.04)

0.64 (p = 0.04)

Time = AM2

 BRACHA score

0.49 (p = 0.06)

0.01 (p = 0.96)

0.29 (p = 0.29)

 PAAS score

0.38 (p = 0.16)

−0.09 (p = 0.75)

−0.18 (p = 0.51)

 Number of aggressive incidents

0.53 (p = 0.04)

−0.10 (p = 0.72)

0.18 (p = 0.52)

 Severity of nearest aggressive incident (n = 10)

−0.32 (p = 0.37)

0.20 (p = 0.57)

0.11 (p = 0.77)

Time = PM

 BRACHA score

−0.14 (p = 0.62)

0.01 (p = 0.97)

0.17 (p = 0.54)

 PAAS score

0.17 (p = 0.55)

−0.14 (p = 0.62)

−0.30 (p = 0.28)

 Number of aggressive incidents

−0.14 (p = 0.63)

−0.06 (p = 0.83)

0.05 (p = 0.87)

 Severity of nearest aggressive incident (n = 9)

0.50 (p = 0.17)

−0.02 (p = 0.96)

0.20 (p = 0.60)

Spearman’s rho values, with p-values in parentheses

Salivary Testosterone and DHEA

We found no differences between the two groups in concentration of testosterone. However, the first morning testosterone concentration correlated significantly with the severity of the nearest aggressive incident (r = 0.64, p < 0.05). The first morning salivary testosterone also tended to correlate with the BRACHA score (r = 0.40, p = 0.11). Although salivary DHEA concentrations did not differ significantly in the two groups, the severity of the nearest aggressive incident and the first morning DHEA concentration were positively correlated (r = 0.64, p < 0.05). Neither hormone correlated significantly with PAAS score.

Discussion

Our pilot data demonstrate the feasibility of collecting saliva from children on an inpatient psychiatric unit, affirm the utility of the BRACHA in predicting violent behavior, and suggest links between salivary hormones and violence. Second morning salivary cortisol concentrations in samples obtained during hospitalization were higher in the children who exhibited aggressive behavior on the units; this finding is even more remarkable because 50 % of these aggressive boys took α2 agonists, which may dampen pituitary-adrenocortical activity and decrease cortisol concentrations in children [10]. We also found that the androgen testosterone and its precursor DHEA were associated with severity of aggression. In the age group studied, these steroids derive mainly from the adrenal glands.

In this data set, the BRACHA-based categorizations of low and high aggression risk groups predicted aggression almost perfectly, making it impossible to know whether salivary cortisol results would outperform, or add incremental information to, BRACHA scores in a larger sample. Although our findings are consistent with prior long-term studies that suggest possible correlations between alterations in cortisol levels and pediatric aggression, this is the first short-term inpatient study for 7- to 9-year-old that assessed cortisol secretion dynamics and absolute concentrations [1113]. The present study’s findings indicate potential usefulness of evaluating cortisol changes over time for predicting and understanding pediatric aggression, a finding that has not previously been documented in this age group.

We note several limitations to our study. First, we did not adjust p-values for multiple comparisons. We feel this approach is acceptable given that this was a pilot study. While the samples were all collected by the same individual (K.A.), samples were not obtained at exactly the same time, and situational factors like eating or drinking were not entirely controlled. We also note that our samples reflected considerable heterogeneity; though all participants were young boys, they had various internalizing and externalizing disorders, and we did not conduct structured diagnostic assessments. We therefore do not know whether the findings described herein might relate to specific forms of psychopathology. Also, we did not collect data about whether aggressive incidents were predatory or impulsive. Future studies should evaluate data relevant to these distinctions as well as data from girls. Lastly, the medications in aggressive and nonaggressive groups differed in that only boys in the former group received α2-adrenergic agonists. Clonidine and guanfacine are prescribed for ADHD and have been associated with significant decreases in cortisol levels [10]. Thus, the average cortisol levels might have been higher and might have differed more significantly had these medications not been prescribed to several boys in the aggressive group. Although other psychotropic medications could affect hormone levels, we found no difference in the use of medications (except clonidine and guanfacine) between the aggressive and non-aggressive groups (Table 2).

In conclusion, our results establish the feasibility of evaluating salivary biomarkers for aggression by children. Follow-up studies in larger samples are required to elucidate whether salivary hormones with the BRACHA scores may improve prediction of aggression beyond what can be established through BRACHA scores alone and the relationships between hormone levels and aggression by children who undergo psychiatric hospitalization. Depending on the outcome of such studies, salivary hormone levels may prove to be useful, practical, non-invasive biomarkers that could form a part of routine testing on child inpatient units, both for anticipating and addressing aggression and for monitoring treatment response. If the results of larger studies are positive, hospitals could improve inpatient safety by inexpensively and rapidly testing salivary hormones to help determine needed observation levels and to aid in selection of effective psychotropic medication.

Acknowledgments

The project described was supported in part (Mr. Blom) by the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant 8 UL1 TR000077-04. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This project was in part supported by NIMH P50 MH077138. The authors gratefully acknowledge funding support from the American Academy of Psychiatry and the Law’s Institute for Education and Research.

Conflict of interest

Jeffrey Strawn declares he has received research support from Eli Lilly, Shire, American Academy of Child & Adolescent Psychiatry and honoraria from the American Academy of Child & Adolescent Psychiatry. Drew Barzman declares he has research support either recently or currently from NIMH, Oxley Foundation, Cincinnati Children’s Hospital Medical Center, American Academy of Psychiatry and the Law Institute for Education and Research, and the Center for Clinical and Translational Science and Training (CCTST) (University of Cincinnati Academic Health Center). Bianca Patel declares that she has received funding from the American Physician Institute in association with CMEtoGO. Dr. Geracioti receives research funding from the U.S. Department of Defense and is a shareholder in RxDino, LLC. Melissa DelBello declares that she has received research support from AstraZeneca, Eli Lilly, Johnson & Johnson, Janssen, Pfizer, Otsuka, Sumitomo, NIDA, NIMH, NIAAA, NARSAD, GlaxoSmithKline, Merck, Novartis, and Lundbeck. She is on the lecture bureau of Bristol-Myers Squibb, Merck, and Otsuka. Dr. DelBello declares that she has consulted, been on the advisory board, and/or has received honoraria from Merck, Schering-Plough, and Pfizer. The other authors declare that they have no conflicts of interest.

Copyright information

© Springer Science+Business Media New York 2013