Presión Profunda en niños con Autismo

Behavioral and Physiological Effects of Deep Pressure on Children With Autism: A Pilot Study Evaluating the Efficacy of Grandin’s Hug Machine

Stephen M. Edelson, Meredyth Goldberg Edelson, David C. R. Kerr, Temple Grandin

Stephen M. Edelson, PhD, is Research Associate, Autism Research Institute, San Diego, California.

Meredyth Goldberg Edelson, PhD. is Associate Professor of Psychology, Willamette University, 900 State Street, Salem , Oregon 97031.

David C. R. Kerr, was Student and Research Assistant, Willamette University, Salem, Oregon, at che time of chis study.

Temple Grandin, PhD, is Assistant Professor of Animal Sciences, Colorado State University, Fort Collins, Colorado.

This article W/JS accepted far publication August 27, 1998


One symptom common to many persons with autism is a high arousal or anxiety leve!. This study investi­gated the effects of deep pressure on arousal and anxiety reduction in autism with Grandin’s Hug Machine, a device that allows selfadministration of lateral body pressure.

Method. Twelve children with autism were randomly assigned to either an experimental group (receiving deep pressure) ora placebo group (not receiving deep pressure but in the disengaged Hug Machine). Al! children received two 20-min sessions a week over a 6-week period. Arousal was measured behaviorally with the Conners Parent Rating Sea/e and physiologically with galvanic skin response (GSR) readings.

Results. Behavioral results indicated a signifi.cant reduction in tension anda marginal/y signiji.cant reduction in anxiety far children who received the deep pressure com­ pared with the children who did not. Additionally, children in the experimental group, whose GSR measures decreased, on average, after deep pressure, were somewhat more likely to have higher GSR arousal a priori.

Conclusion. These preliminary jindings support the hypothesis that deep pressure may have a calming ejfect far persons with autism, especial/y those with high levels of arousal or anxiety.

Persons with autisrn are ofcen described as having rel­ atively high arousal or anxiety levels (Hardy, 1990; Sands & Ratey, 1986; Wing, 1989). Hutt, Hutt, Lee, and Ounsted (I 965) noted chat children with autisrn dernonstrate a desynchronized electroencephalogram pat­ tern, indicating high levels of arousal. This desynchroniza­ tion was relaced to increased environrnental  stirnulation and increased stereotyped (i.e., repetitive) behavior.

The underlying reason for che high levels of arousal in sorne persons with autisrn is not entirely known. Sorne researchers suggested that a high leve! of arousal rnay be due to neurological dysfunction, such as sensory processing problerns (Delacato, 1974; Grandin, 1995); others posited chat che problern relates to faulty inforrnacion processing (Ornitz, 1985); and ochers suggested chat a rnyelination defect mighc lead to increased neuronal arousal (McClel­ land, Eyre, Watson, Calven, & Sherrard, 1992).

One rnethod clinicians use to lessen  anxiety  and arousal in persons with autisrn is che applicacion of deep pressure. For example, Ayres (1979) and King (1989) reported that wrapping a child with autism in a gym mat produces a calrning effect. Persons wich aurisrn also have been known to provide thernselves with deep pressure in an attempt  to  calrn  themselves  (Grandin,  1992; & Scariano, 1986) and often prefer to provide this stimula­ tion themselves, frequently avoiding tactile stimulation controlled by others (Delacato, 1974).

There has been little empirical research on the efficacy of deep pressure on persons with autism. lnamura,  Wiss, and Parham (1990a, 1990b) investigated the effects of Grandin’s Hug Machine, a device that provides deep pres­ sure to the lateral parts of the body, on the behavior of nine children with autism. Temple Grandin, a woman with autism, developed the Hug Machine after observing that cattle showed a dramatic decrease in anxiety when receiv­ ing deep pressure from a cattle chute, a device used during branding. Grandin modeled the Hug  Machine  on  the cat­ tle chute but modified the design for  human  use (Grandin & Scariano, 1986). She proposed that persons with rela­ tively high levels of anxiety or arousal are more likely to benefit from the Hug Machine  than  those  with  moderare to low levels of anxiety or arousal.

lnamura et al. (1990a, 1990b) noted that, in general, the children in their study did not consistently use the Hug Machine, with duration of use being less than 2 min per session. However, greater use seemed to be related to decreased hyperactive behavior in sorne children. Although these findings are encouraging, lnamura et al. did not use a control group, did not use physiological measures to see whether there were concomitant physio­ logical changes due to deep pressure, and did not analyze their data statistically.

In an open clinical study, Creedon (1994) found that children with autism who used the Hug Machine for longer periods and with more sustained pressure on days that were associated with behavior problems made fewer aimless actions, made more adaptive movements, and were able to sit more calmly than children who did not use the Hug Machine regularly. Similar to Inamura et al.’s (1990a, 1990b) study, Creedon’s report was based on clinical rather than statistical data.

Another type ofintervention involving deep pressure is holding therapy (Welch, 1988). This treatment involves holding the person for long periods while providing deep pressure. Because the person does not control the deep pressure, one criticism of holding therapy is that he or she may not desire the pressure or the amount of pressure applied. Additionally, children often resist the initial hold­ ing therapy sessions, which can lead to considerable stress for the child. Therefore, the calming effects noted by Welch may be due to learned helplessness rather than the tactile stimulation. We are unaware of scientific studies assessing the efficacy of this intervention.

Field et al. (1997) investigated the use of touch therapy on 22 preschool children with autism, half of whom received touch treatment and the other half a placebo intervention. The form of touch therapy these researchers used involved providing “moderare pressure and smooth stroking movements on each of the following areas: head/neck, arms/ hands, torso, and legs/feet” (p. 334). The results indicated increases in attention to multiple tasks, social and relating behavior, and initiating behavior, and a decrease in general sensory problems. lt should be noted that although touch therapy provides tactile stimulation, it is not deep pressure.

Two case studies reported dramatic behavioral changes as a result of other forms of tactile stimulation that more closely resemble deep pressure. In one study, foam arm splints were placed on a child with autism, resulting in a decrease in self-injurious behavior and an increase in social interaction (McClure & Holtz-Yotz, 1991). In  another study, a reduction in stereotypic, self-stimulatory behaviors was observed after frequent back rubs and hugs from a ther­ apist (Zissermann, 1992). When the child also wore tight­ fitting gloves and a weighted vest, which provided contin­ uous pressure, stereotypic behaviors and hand  slapping were further reduced. As clinical reports, these case studies did not empirically test the efficacy of deep pressure.

Finally, Krauss (1987) investigated the effects of deep pressure on college students using a device called the Hug’m Apparatus. Anxiety level was determined with a physiological measure (heart rate) and a self-report measure (State-Trait Anxiety Inventory). Although there was no sig­ nificant change in heart rate as a result of the deep pressure, the State-Trait Anxiety lnventory indicated a greater re­ duction in anxiety levels in the experimental group than in the control group. However, this difference was not signif­ icant. Krauss speculated that initial low levels of state anx­ iety may, in part, explain the limited change in anxiety after deep pressure noted in the experimental group.

The aim of the present study was to provide a more controlled empirical investigation of the effects of deep pressure, using Grandin’s Hug Machine, on children with autism. In contrast to Inamura et  al.’s  (1990a,  19906) study, we used an experimental group receiving deep pres­ sure and a matched (a priori on observable indicators of anxiety, such as jitteriness, shakiness, etc.) placebo group that did not. Unlike the clinical reports ofCreedon (1994), McClure and Holtz-Yotz (1991), and Zissermann  (1992), we specifically wished to conduct statistical comparisons of children who did and did not receive deep pressure to see whether any observed changes would meet the criteria of significance. Finally, similar to Krauss’s (1987) study with college students, we wanted to investigare the relationship between behavioral and physiological indicators of arousal before and after deep pressure in children with autism.

The specific goals of the present study were to deter­ mine whether (a) deep pressure affected behavioral indexes of anxiety, (b) deep pressure affected physiological indexes of anxiety; and (c) there were unintended side effects of deep pressure. lt should be noted that because of our small sample size, our investigation should be considered a pilot study, an initial attempt to answer these questions.



Founeen children who were diagnosed with autism by a physician were recruited from the Portland and Salem areas of Oregon. Two children were excused from the study (both boys) because one did not want to be separated from his parents during the experiment and the other would not allow us to obtain galvanic skin response (GSR) measures. Therefore, our final sample induded 12 children (9 boys, 3 girls), ranging in age from 4 to 13 years (M = 7.58 years, SD = 2.91). Half of the children had meaningful commu­ nication skills, and half were either nonverbal or verbally impaired (e.g., echolalic).

If possible, the children were matched on the basis of age and gender and two of the authors’ independent evalu­ ations of anxiety. In all cases, the two authors were in inde­ pendent agreement about the marches. After the marches were formed, one member of each match was randomly assigned to the experimental condition, the other to the placebo condition. Unfortunately, both children in one match were mistakenly run in the placebo condition. Thus, there were 5 children in the experimental group and 7 in the placebo group. There was no statistical difference between the children’s ages in either group.


Hug Machine. The Hug Machine (see Figure 1) is con­ structed of two padded side boards that are hinged near the bottom to form a V-shape. To use the device, the person líes down or squats between the two side boards. By pulling a lever, the user engages an air cylinder that pulls the boards together. This action provides deep pressure stimulation evenly across the lateral parts of the body. Given lnamura et al.’s (1990a, 1990b) repon about the limited use of the Hug Machine by their sample, we encouraged the children in our study to use the device often and helped them to do so, if necessary.

Galvanic skin respome. The GSR was used to examine changes in physiological arousal. Although a high arousal level does not necessary imply a high anxiety level, the two variables are highly correlated (Alexander, White, & Wallace, 1977). The GSR has also been used in other research studies to measure the anxiety of persons with autism (Bernal & Miller, 1971;  Stevens  &  Gruzelier, 1984). The Temperature and  Skin  Conductance  Kit (Model 201T) 1 was used to measure GSR

Conners Parent Rating Sea/e (CPRS). Behavioral indica­

tors of anxiety were assessed with the CPRS (Conners, 1970; Goyette, Conners, & Ulrich, 1978). The CPRS con­ tains 93  items  relating to numerous  behaviors, induding social, anxiety, compliance, obsessiveness-compulsiveness, and hyperactivity, but we were only interested in the items tapping different dimensions of anxiety. Therefore, we formed three anxiety scales by creating linear combinations of questions assessing anxiety-related behaviors. The Anxiety scale examined general indicators of anxiety, induding overall physical anxiety, fear, and excitability. 1t induded the following items from the CPRS: afraid of peo­ ple, afraid of being alone, restless or overactive, excitable or impulsive, cannot stand too much excitement,  unable  to stop a repetitive activity, and acts as if driven by a motor. The Tension scale induded ali the items on the Muscular Tension section of the CPRS: gets still and rigid; twitches, jerks, etc.; and shakes. The Restlessness-Hyperactivity scale induded the following CPRS items: restless; fails to finish things he or she starts (short attention span) and  inatten­ tive, easily distracted (both of which are also on  the gener­ al anxiety scale); and constantly fidgeting.

The range of possible seores for each ítem was 1 to 4 (1 = the problem behavior  bothered  the child  not at ali, 2 =justa little, 3 = pretty much, 4 = very much). Thus, the total score could range from 7 to 28 for the Anxiety scale,

3 to 12 for the Tension scale, and 6 to 24 for the Rest­ lessness-Hyperactivity scale.

Side ejfects questionnaire.The side effects questionnaire

was developed for the study and used primarily as a mea­ sure of possible side effects of the deep pressure. At the beginning of each week, the children’s parents were asked to complete the two-item, open-ended questionnaire to repon whether they had noticed subtle or dramatic changes in their child’s behavior that they could attribute directly to their child being in the Hug Machine.


Before the study, each child was placed in the Hug Ma­ chine for one session to familiarize him or her with the lab­ oratory room, the GSR electrodes, and the deep pressure device. A1l the children spent 5 min to 1O min in the device and were encouraged to actívate the deep pressure with the lever to ensure that they could tolerare the deep pressure should they be assigned to the experimental condition. A few children who were hesitant to go into the Hug Machine were brought back for a second time to familiar­ ize them again with the device and with the experimental procedures. After this “pretraining,” the children received 12, 20-min sessions in the Hug Machine (i.e., twice a week for 6 consecutive weeks).

Children in the experimental group were instructed to use the Hug Machine lever to provide deep  pressure  as often as they desired. Children in the  placebo  group also lay in the Hug Machine, bue the lever was disengaged so that they were not able to provide themselves with deep pressure. In addition, the side boards were moved apart so that deep pressure could not be provided manually.

Therefore, the only pressure provided to the placebo group was that of the child’s body against the floor.

GSR was measured before and immediately after each session. Probes were coated with electrode gel and attached to the index and middle fingers of the child’s right hand with hook-and-loop tape. The GSR readings began 15 sec after the electrodes were attached. Three recordings were made during the 15-sec time interval: (a) the GSR reading at the beginning of the interval (GSR-15), (b) the maxi­ mum GSR reading within the time interval (GSR-max), and (e) the minimum GSR reading within the time inter­ val (GSR-min).

Parents were instructed to complete the CPRS before the 1st session, after the 6th session, and after the 12th session. Because parents were blind to group assignment, they waited in a room adjacent to the laboratory  while  their child was in the device. In addition, they were not present for the GSR measurements, which were conducted imme­ diately before the children entered the Hug Machine and immediately after they exited the Hug Machine.


lntercorre/,atiom of Behavioral and Physiological Data Befare Deep Pressure

Table 1 presents the results of an initial correlational analy­ sis to determine the degree of relationship between the behavioral and physiological indexes of anxiety or arousal before the experimental conditions. As can be seen, these variables were well correlated. Pearson correlations coeffi­ cients among GSR-min, GSR-max, and the three CPRS scales ranged from .46 to .65. These were ali at least mar­ ginally significant despite the small sample size. Moreover, the two GSR variables were highly correlaced with one another, r(l2) = .98, p < .001. Similarly, the three CPRS scales were well correlated with each other. The correlacion between the Anxiety and Tension scales was r(l2) = .80, p < .O l; the correlation between the Anxiety and Rest­ lessness-Hyperactivity scales was r(l2) = .90, p < .001; and the correlacion between che Tension and Restlessness­ Hyperactivity scales was r(I 2) = .64, p < .05.

These results indicare that behavioral and physiologi­ cal indicacors of arousal converged. They also validare che usefulness of the three anxiety scales that we created from the CPRS. lnterestingly, although the physiological vari­ ables remained highly correlated with each other as did the behavioral variables throughout the course of the study, the intercorrelations between the behavioral and physiological indicators of arousal were no longer significanc either at the midpoint or the end of the study.


Although chis muse be considered a pilot study because of che relatively small sample size, che behavioral results seem to support che contention that deep pressure has a calming effect on children who are anxious. The children who received deep pressure demonstrated a significant decrease on che Tension scale anda marginally significant decrease on the more general Anxiety scale.

The findings from che GSR measures furcher suggest that five children in the experimental condition seemed to benefit from deep pressure sessions. These benefitors were chose who had che highest inicial levels of physiological arousal. Ayres and Tick.le (1980) noted chat sensory inte­ gration therapy was also more effective for children with autism who had normal or overaroused sensory responsive­ ness chan for children whose systems were underaroused. Thus, interventions directed at changing che sensory sys­ tem might be more efficacious for persons wich higher lev­ els of arousal. Anecdocally, che child who had che highest a priori GSR leve! in our study and had been assigned, at random, to che experimental group showed che most dra­ matic decrease in arousal over che 12-session intervention period.

lnterestingly, the concordance between che behavioral and physiological indicators of arousal, noted before che children were in eicher Hug Machine condition, did not remain throughout the course of the study. The reason for this may be that sorne children improved behaviorally but not physiologically. Furthermore, the small sample size may have either sufficiently reduced the stapstical power of our analyses or weakened the correlations between the behavioral and physiological measures. lt should be noted that even though the correlations between these two class­ es of variables were no longer significant by the midpoint and end of the study, the pattern of positive correlations was largely unchanged.

lt seems possible that the behavioral CPRS data reflect­ ed the positive effects of the Hug Machine, whereas the physiological GSR data reflected either discomfort with the skin probes or the stress involved in having the GSR mea­ sured. Van Engeland, Roelofs, Verbaten, and Slangen (1991) reported an inability to distinguish hyperarousal in children with autism from their reactions to the stress of being tested. Moreover, James and Barry (1984) found abnormal autonomic reactions to repeated stimulation of persons with autism. lt is possible that subjecting the chil­ dren to repeat!!d GSR measurements resulted in increas­ ingly abnormal physiological reactions over time. This might also explain the failure to find many physiological changes in response to the Hug Machine.

This study has several limitations. Although we tried to recruit children who were highly anxious, the majority had moderare levels of anxiety. Future studies rnight screen participants for anxiety level a priori to further explore the threshold hypothesis that persons with high levels of anxi­ ety will benefit the most from the Hug Machine. Additionally, the children’s use of the Hug Machine was arbitrarily scheduled for two sessions a week for 6 weeks, yet Grandin was able to use the Hug Machine during times when she was most anxious (Grandin & Scariano, 1986). lt is not known whether the eflicacy of the Hug Machine was reduced by scheduling arbitrary sessions rather than by allowing the children to have access to the device during periods of heightened arousal or anxiety. lnamura et al. (1990a, 1990b) noted that greater Hug Machine use seemed to produce greater positive effects for sorne of the children in their study, and Creedon (1994) noticed that children with autism used the Hug Machine more and for a greater duration on the days they displayed more behav­ ioral problems. Therefore, increasing access to the Hug Machine as well as allowing for its use during self-deter­ mined times may possibly increase its efficacy. Although it was not feasible to test this possibility in our study, ideally the Hug Machine would be more accessible when used. clinically.

As already stated, another limitation was the small sample size, which greatly reduced the power of the statis­ tical analyses. The final limitation was the nature of our sample. Because half the sample was functionally noncom­ municative, we were not able to obtain self-report data regarding these children’s subjectively experienced anxiety or arousal. Self-reports would have provided further insight into their feelings before, during, and after the deep pres­ sure sessions. Because of the communication impairments, it is not known whether deep pressure had any effect on these children’s subjective feelings of anxiety or arousal as would be expected from Krauss’s (1987) findings.


The results of this pilot study are encouraging. Deep pres­sure appears beneficial for children with high levels of anx­ iety or arousal, and there may be a threshold of anxiety or arousal required for deep pressure to be  beneficial. Thus, our research supports the reports of clinicians  and  parents of children with autism who have noted the calming effects of deep pressure for this population (Ayres, 1979; Creedon, 1994; King, 1989; McClure & Holtz-Yotz, 1991; Zisser­ mann, 1992). Because of the limitations noted, more research is needed to replicare and extend our results.