When people become stressed, their bodies can respond by sweating.
Now, researchers at the University of Missouri are monitoring how much adolescents severely affected by autism sweat in order to better understand when behavioral issues, such as aggression, are likely to occur.
Bradley Ferguson analyzed the stress levels of eight adolescents who are severely affected by autism spectrum disorder at The Center for Discovery, a residential facility in New York that provides advanced care and research for individuals with complex conditions. Using wrist and ankle monitors, Ferguson found that there was a rise in the body’s electrodermal activity – which results from increased levels of sweat – 60% of the time before an individual showed behavioral issues.
“A spike in electrodermal activity is telling us that the individual’s body is reacting physiologically to something that is stressful, which could be their internal state, something in the environment, or a combination of the two,” said Ferguson, assistant research professor in the departments of health psychology, radiology and the Thompson Center for Autism and Neurodevelopmental Disorders.
“If parents or caregivers are notified ahead of time that their child’s stress levels are rising, they might have a chance to intervene and de-escalate the situation before problem behaviors occur.”
Ferguson explained that possible intervention methods could include removing the child from the environment or activity that is causing the stress, as well as providing access to an item that the child enjoys interacting with in an effort to calm them.
Credit: Mizzou News.
“Individuals who are severely affected by autism spectrum disorder are often unable to verbally communicate their discomfort when they become stressed,” Ferguson said.
“However, their body still responds to stressors just like anyone else.
Therefore, being alerted of increases in electrodermal activity can allow parents and caregivers to intervene prior to engagement in problem behavior with the goal of ensuring the health and safety of those involved.”
Ferguson collaborated on the study with David Beversdorf, a professor of radiology, neurology and psychology in the MU College of Arts and Science as well as principal investigator of the Cognitive Neuroscience Laboratory in the MU School of Medicine. Ferguson also collaborated with Theresa Hamlin, Johanna Lantz, and Tania Villavicencio at The Center for Discovery, and John Coles at Calspan-University of Buffalo Research Center and The State University of New York at Buffalo.
Ferguson explained that possible intervention methods could include removing the child from the environment or activity that is causing the stress, as well as providing access to an item that the child enjoys interacting with in an effort to calm them.
“Important work is being done to try to identify predictors for when a person with autism is at greatest risk of having a behavioral episode,” Beversdorf said.
“This research highlights the individual variability in this response that must be considered, and may also have implications for individualized treatment approaches moving forward.”
Funding: The study was funded by the New York State Center of Excellence, New York State Department of Health and Office for People with Developmental Disabilities, as well as private monies donated to The Center for Discovery. The content is solely the responsibilities of the authors and does not necessarily represent the official views of the funding agencies.
The Department of Health Psychology is in the MU School of Health Professions, and the Department of Radiology is in the MU School of Medicine.
Autism spectrum disorder (ASD) is a neurodevelopmental condition that is characterized by persistent deficits in social communication and restricted, repetitive patterns of behavior that occur early in development (1).
Research has demonstrated that many individuals with ASD engage in aggression and other problem behaviors such as self-injurious behavior and irritability. For example, one study found that in a sample of 1,380 children with ASD, up to two-thirds became aggressive toward their caregivers, and nearly half became aggressive toward others (2).
Those individuals with ASD with greater intellectual impairment, often with significantly limited communication and social engagement skills, are at greater risk for the development of problem behaviors (3).
Problem behavior poses many challenges for caregivers and service providers alike, with the potential to become worse over time (4). In order to attempt to reduce the occurrence of problem behavior, research has begun to explore the use of psychophysiological markers preceding problem behavior so that an intervention may occur prior to the onset of the behavior.
Among those with ASD, high levels of stress may manifest as problem behaviour (5). Additionally, emotional regulation appears to be impacted in general in ASD (6).
Knowing the physiological stress state of individuals with ASD may allow caretakers and service providers to intervene prior to the occurrence of problem behavior. Furthermore, better understanding environmental correlates of physiological responses can help to develop more precise treatments through control of or manipulation of these factors. One psychophysiological technique to assess an individuals’ internal stress state is the measurement of electrodermal activity (EDA).
Increases in EDA indicate activation of the sympathetic nervous system (the “fight-or-flight” response of the autonomic nervous system), which is measured by assessing changes in electrical conductivity between two electrodes placed in close proximity to each other on the skin.
Activation of the sympathetic nervous system results in secretion of sweat, which conducts electricity, from eccrine sweat glands throughout the body. Eccrine sweat glands are only innervated by the sympathetic branch of the autonomic nervous system, so increases in EDA can be, in part, attributed to increases in physiological arousal. EDA is typically measured from areas of the body with a high density of eccrine sweat glands, such as the palms of the hands or soles of the feet, but can also reliably measure sweat secretion when placed on other areas of the body such as the wrist or immediately above ankle (7). Measurement of EDA has been shown to be well-tolerated in ASD and is sensitive to changes in arousal and emotional states in this population (8–10). However, changes in EDA in response to an arousing stimulus, vary widely in ASD.
For example, abnormal baseline EDA as well as both hypo- and hyperactivities in response to human faces have been shown in ASD (11). Typically, EDA returns to baseline shortly after the application of an arousing stimulus. However, some children with ASD fail to return to their baseline EDA after the occurrence of an environmental stimulus (11), suggesting that a large stress response may continue to affect behavior long after the occurrence of a stressful event, including the engagement in problem behavior. A recent study found a relationship between EDA and externalizing behavior problems during compliance-oriented play tasks (12).
However, EDA was relatively low while the individual engaged in the problem behavior, suggesting heterogeneity in the autonomic nervous system response to task demands in ASD. Furthermore, greater variability in EDA in response to a battery of naturalistic and structured parent–child, child alone, and direct testing tasks has been shown to be associated with overall ASD severity (13), suggesting that EDA may be a promising predictor of problem behavior in ASD.
In light of this, research is now beginning to explore the utility of psychophysiological markers to anticipate problem behaviors in ASD. One recent study has utilized multimodal psychophysiological arousal to successfully predict aggression in minimally verbal ASD patients in a naturalistic setting (14).
A great amount of heterogeneity exists in ASD, which can complicate research efforts when examining data at the group level. As such, identification of subtypes within ASD may lead to more effective treatments (15, 16).
Psychophysiological data has recently been used to identify relationships between co-occurring symptoms in ASD, such as gastrointestinal symptoms, irritability, and sleep problems (8).
Therefore, while psychophysiological data may also be useful to assess the internal stress state of individuals with ASD in a variety of settings, it is important to examine the heterogeneity of this response. Understanding this variability will be important in the development of tools to anticipate the onset of behaviors, which may help lead to more individualized treatment approaches. This will be of a particularly acute need in the most severely affected individuals, for whom problem behaviors are more frequent (3).
The present study first examined the feasibility of collecting EDA data from individuals with severe ASD in a naturalistic setting, while participating in skill acquisition in a school setting at a residential facility for severely affected individuals with developmental disabilities.
The lab school at The Center for Discovery (TCFD) utilizes discreetly mounted video cameras and microphones in classrooms to collect behavioral data, while students wear physiological data collection sensors to gain a better understanding of the physiological correlates associated with learning and behavior in individuals with ASD. As such, we wished to examine the feasibility of examining psychophysiological variables in individuals with severe ASD as they are related to problem behavior.
The beginning and end of problem behaviors were identified through video recordings and were confirmed by trained staff at TCFD. The associated EDA (i.e., time locked to the EDA recordings) was monitored prior to the occurrence of the initial problem behaviors as well as immediately after the cessation of the problem behavior, to examine individual variability in this particular psychophysiological variable as it related to problem behavior.
Results
Frequency of Problem Behaviors
To demonstrate the range of behaviors among the students, behavior frequency each participant engaged in over the year-long period assessed is depicted in Figure 1. Overall, most of the 8 students engaged in more than one type of behavior, except for student 11 who only engaged in self-injurious behavior. For all of the documented behaviors across all of the students, the most common problem behavior was self-injurious behavior, followed by aggression, inappropriate social behavior, non-compliance, and elopement.
(A) Frequency counts of the types of problem behaviors engaged in over 1 year for each student in the sample. (B) Total number of behaviors engaged in by the sample over 1 year and proportion of the whole that each behavior represents.
Quality of EDA Signal
Initially, the EDA data were visually inspected by an experienced psychophysiology researcher (BF) for artifacts as the Q-sensors were placed either on the wrist or the ankle of the students. Any cases with significant artifacts in EDA during the times that were analyzed, as defined by a visual analysis of motion from accelerometer data in the x, y, and z directions at the same time as significant EDA artifacts were noted, or DC shift, indicating either loose or no contact between the skin and the Q-sensor (25), were excluded from the final analysis. Significant EDA artifacts were defined as an immediate drop in EDA to 0, indicating DC shift, or when “spikes” or rapid increases in EDA data appeared and were at least three standard deviations above the mean. This typically corresponded with rapid fluctuations in accelerometer data in x, y, and z directions and was present during periods when the student engaged in problem behavior that resulted in movement of the arms and/or legs. Artifacts in the EDA data were easily detected as they tended to be characterized by quick “spikes” in the EDA data that aren’t physiologically likely, followed by DC shift, indicating that contact between the EDA sensor and the skin was broken. Visual inspection of each EDA record yielded 22 records that contained a significant amount of artifact, according to the visual analysis mentioned above, that rendered the data unreliable, and so they were excluded from the final analysis. Each of these 22 records were associated with participant behaviors. This yielded a total of 62 valid records that were suitable for analysis.
Anticipatory Rise Time Prior to Problem Behavior
The presence or absence of an anticipatory rise in EDA prior to a student engaging in problem behavior was analyzed. On average, across all episodes of problem behavior documented, across all students, the students displayed an anticipatory rise in EDA prior to engaging in problem behavior 60% of the time. However, individuals varied in the frequency with which an anticipatory rise in EDA was observed before the onset of the problem behavior (see Table 2). Of note, the students with the greatest incidence of anticipatory rise in EDA had primarily engaged in general classroom disruption behaviors and self-injurious behavior, while the student with the least amount of times of an anticipatory rise had primarily engaged in aggression (see Table 2).
Recovery of EDA to Baseline After Problem Behavior
On average, across all problem behaviors documented, the students’ EDA returned to median baseline values 45% of the time after engaging in problem behavior. Individuals also varied in the frequency with which EDA returned to median baseline levels after cessation of the problem behavior. Of note, the student who returned the most frequently to their median baseline EDA primarily engaged in out-of-seat behavior, while the student who returned the least amount of times to their median baseline EDA primarily displayed agitation (see Table 2).
Examples of the different types of EDA responses are illustrated in Figure 2. For some problem behavior, the EDA pattern was characterized by a gradual increase in EDA leading up to the problem behavior, a peak in EDA while the individual was engaged in the problem behavior, and then a gradual decrease in EDA following discontinuation of the problem behavior. For some problem behavior, the EDA pattern was characterized by no build-up in EDA prior to the engagement of the problem behavior, with a gradual increase in EDA during engagement in the problem behavior, followed by a decrease in EDA following discontinuation in the behavior. For other episodes of problem behavior, the EDA pattern was characterized by an increase in EDA prior to engagement in the problem behavior which was followed by a decrease in EDA during engagement of the problem behavior (see Figure 2 for examples of each response pattern).
Examples of skin conductance level traces by behavioral subtype. For all images, the blue-shaded area represents the time period of an individual engaging in problem behavior. The left-most blue pin represents the beginning of the problem behavior, and the right-most pin represents the end of the problem behavior. (A) No EDA build-up prior to engaging in problem behavior (in this example, jumping in seat); (B) EDA build-up prior to engaging in problem behavior (in this example, jumping in seat); (C) EDA build-up prior to engaging in problem behavior (in this example, repetitive body hitting) and subsequent reduction with engagement in problem behavior. The x-axis is EDA in microsiemens, and the y-axis is time (marked at the top of each EDA record).
Discussion
The findings from this exploratory study suggest that examining the relationship between EDA and problem behaviors is feasible in a naturalistic setting in a severely affected population with ASD. There was some loss of data due to artifact, which was expected due to the level of functioning of the students and the assessment of EDA prior to and during the engagement of problem behavior that involves motion, rather than in a well-controlled laboratory setting. However, given the exploratory nature of the study, the findings suggest that the collection of EDA in a population with severe ASD is possible, but investigators should be aware of a number of limitations that are likely to occur in this population.
Variability was found in the incidence of changes in EDA prior to the occurrence of problem behavior, during the problem behavior, and after the occurrence of a problem behavior, targeting a unique set of severely affected individuals in a residential setting. The results indicate that 60% of the total episodes of problem behavior documented were associated with an anticipatory rise in EDA prior to engaging in problem behavior. This has important implications for efforts to utilize psychophysiological markers as a predictor of problem behaviors (14). This finding could also have significant implications for the prediction and management of problem behavior in a classroom setting of individuals who are severely affected by ASD. For those cases where an anticipatory rise was found, the average amount of time that EDA rose prior to engaging in problem behavior was over 10 min, providing a window of opportunity for intervention to occur prior to the occurrence of problem behavior. However, this would not be helpful in all cases, as many behaviors were not associated with an anticipatory rise in EDA
. Further monitoring of these individuals may be helpful to understand other factors that might predict behaviors, and in how many of these individuals do other psychophysiological variables contribute predictive information. Given the variability in behavioral profiles of individuals with ASD in this study, future work will need to determine whether specific behavioral profiles are associated with certain patterns of EDA change, allowing improved specificity for the psychophysical prediction efforts. Many have suggested that understanding the psychophysiological underpinnings of problem behavior in ASD is important for predicting when an individual with ASD is likely to engage in problem behaviour (12, 26), but the distinct patterns of EDA suggest that different treatment strategies may be more effective for each type of EDA response pattern. For example, individuals with ASD that display a steady increase in EDA prior to engaging in problem behavior may respond best from cognitive and/or behavioral interventions to reduce anxiety.
Additionally, pharmacological treatment with agents targeting stress reactivity may also provide benefit in this subtype. For example, propranolol, an agent utilized in other conditions associated with altered stress reactivity (27), has been shown to increase conversational reciprocity in ASD (9), making it a potential candidate drug for the treatment of anxiety-related behaviors in ASD. However, those without an increase in EDA prior to engaging in problem behavior may benefit from behavioral or pharmacological interventions targeting impulse control.
In this sample, we also identified that over half of the students did not reliably return to their initial resting baseline EDA after engaging in problem behavior. This suggests that, in future work examining the relationship between EDA and behaviors, it will be critical to account for the potential confound of a lack of returning to baseline from previous behaviors for EDA, and which behavioral profiles are associated with lack of return to baseline. Additionally, this raises the possibility that interventions after the cessation of behavior might be beneficial in some individuals, targeting de-escalation.
A number of limitations to the present study should be noted, as they affect the generalization of the results across a broad range of individuals with ASD. First, the students in this study ranged in age from 13 to 20 years, with an average age of 16, were all male and have severe ASD.
As such, it is not clear how this generalizes to all individuals with ASD, and so future studies will need to examine data from a more diverse sample of individuals. Second, detailed data from only eight individuals were analyzed for this study, and so future research should aim to examine these results in a much larger group of individuals with ASD, which would also allow the examination of the impact of co-occurring conditions and medications on the EDA/problem behavior relationship. However, this remains of interest for the management of those with the most severe problem behaviors, who were studied herein, and the variability that will be important in future psychophysiological monitoring in this type of setting.
Finally, when planning psychophysiological experiments in those with severe ASD, careful consideration should be given to the testing environment, the problem behaviors engaged in by the individual(s) to be studied, and the tolerance of biosensors on the individual. For instance, if data collection will be indoors and outdoors, EDA would not be an appropriate measure given that EDA readings can be influenced by changes in hydration status, relative humidity, or sweat, for example (23, 28, 29).
In this case, it may be better to collect electrocardiogram (ECG) to analyze heart rate variability (HRV), which provides information about sympathetic as well as parasympathetic nervous system functioning. Further, if an individual engages in arm flapping or elopement—for example, collection of data from the wrist or leg may not be appropriate given the high probability that the data will contain motion artifact. To this point, EDA data from student 13 in this study was unable to be analyzed due to significant motion artifact from repetitive motor movements (Table 2).
In this case, investigators may consider the use of a physiological apparatus that is affixed to the trunk of the body that is less susceptible to motion artifact. As such, there are a number of limitations of collecting EDA data from those with severe ASD, but with careful planning, such studies are possible and add a wealth of knowledge on those more severely affected.
Source:
University of Missouri-Columbia
Media Contacts:
Brian Consiglio – University of Missouri-Columbia
Image Source:
The image is in the public domain.
Original Research: Open access
“Examining the Association Between Electrodermal Activity and Problem Behavior in Severe Autism Spectrum Disorder: A Feasibility Study”. Bradley Ferguson et al.
Frontiers in Psychiatry doi:10.3389/fpsyt.2019.00654.
