H. Kaplan, Ph.D., M. Clopton, OTR/L, L. Messbauer, OTR/L, M. Kaplan, Ph.D., OTR/L, A. Rodriguez, A.A.S., R. Avigdor, M.S., S. Goldglanz, M.S. & S. Weinberg, M.S.
These experiments explored whether behavioral improvements observed during Snoezelen OT treatment sessions generalized beyond the session to two different settings for a person with moderate intellectual disability, autism and severe challenging behaviors. Experiment 1 explored engagement during a functional task immediately following the treatment sessions while Experiment 2 explored changes in the frequencies of two challenging behaviors on the days after treatment sessions. Investigators found generalization to both post-session engagement as well as to the daily frequency of two challenging behaviors on days following the OT sessions
Snoezelen and other types of multi-sensory environments have been set up at many agencies providing services to people with intellectual disabilities. A literature about their effects has only begun to develop slowly in the last few years (Hogg, Cavet, Lambe, & Smeddle, 2001). Before this, empirical support was lacking and most of the literature was limited to anecdotal descriptions or case studies using few experimental control procedures. (Hogg et al., 2001). A recent study by Cuvo et al. (2001) demonstrated that these environments reduce stereotypical behaviors for some individuals and increase their task engagement while they are in the room. The article, though, raised questions about the extent to which these effects generalized to post-session activities as well as to whether they compared favorably with other readily available and less expensive activities like going for a walk outside.
This pilot study attempted to determine if the current methodology could be used to extend the literature. Our research team investigated whether observed changes in engagement generalized to a post-session functional activity and whether there was further generalization as measured by the frequency of challenging behaviors during the hours/days following the Snoezelen session.
The setting for this series of two pilot experiments was a Day Habilitation program for persons with intellectual disabilities and severe challenging behaviors. We operationally defined a severe challenging behavior as one which has resulted in the person's rejection from at least three other conventional day program settings. Many of the people served in this Day Habilitation program are people with Autistic Spectrum Disorders who were previously at home or institutionalized for extended periods of time. The agency providing these services offers a wide range of O.T. interventions including Snoezelen based services, in several of its other sites and has had anecdotal success using this modality with other individuals exhibiting similar functional profiles. This Day Habilitation program differs from others in that there is a much higher staff to consumer ratio, more space per individual and smaller group sizes, allowing it to be more person centered and flexible in its approaches to assisting consumers in the process of integrating into their local communities.
Participant in this series of pilot experiments was a 35 year old adult male with Autism, Moderate Mental Retardation and Rapid Cycling Bipolar Disorder. At the time of the experiment, he had been attending the above Day Habilitation program for individuals with severe challenging behaviors for the previous six years. His challenging behaviors were crying incidents, cigarette butt pica, elopement and self-injury. He was taking the psychotropic medications Buspar and Risperdol. He had been living in a group home for autistic adults for several years prior to and all through the period of time data were being collected. He was on a behavior plan for the above challenging behaviors which remained constant for the entire time data was being collected during the study, as did his level and type of psychotropic medications. Functional analysis of his challenging behaviors indicated that he was primarily seeking tangible reinforcers, (coffee and cigarettes). He was non-verbal and used sign language, gestures and eye movements to communicate his needs and desires. He related mainly to staff rather than other consumers in the program, most likely because it was staff that usually gave him access to preferred reinforcers. Prior to participating in the experiment, he was not receiving any Occupational Therapy services. The Occupational Therapist remained the same throughout the experiment.
The experimental design used in this pilot experiment was an alternating treatment or reversal design (Cook & Campbell, 1979), also referred to as an ABA design (Kazdin, 1982). In this design, the participant received Occupational Therapy services in a Snoezelen room for the first phase of the study lasting several sessions, then received non-Snoezelen Occupational Therapy for several sessions with a final reversal to Snoezelen Occupational Therapy again for several sessions. Hence, the A B A sequence, where Snoezelen was the "A" phase and Occupational Therapy without using Snoezelen was the "B" phase. If there was some effect of the use of Snoezelen room, then there should be some discontinuity between phases "A" and "B" in the data, and then another discontinuity between phase "B" and "A" at the end of the experiment. Further, the data should trend in the same direction during the two "A" phases and should trend in the opposite direction during the "B" phase. Sessions occurred twice per week.
Experiment 1 investigated whether generalization from the Snoezelen session took place during a five minute sandwich making and eating task immediately after the end of the OT session. This was a functional goal for him at both the day program as well as at the residence and he had extensive experience with the task, decreasing potential threats to internal validity due to learning. He was videotaped during this task and the number of prompts required for him to complete the task was recorded.
Operational Definition of the Dependent Measure:
The Dependent measure was the number of prompts required to complete the sandwich making and eating task. For the purpose of Experiment 1, "prompt" was operationally defined as any gesture or vocalization which cued the participant to remain engaged in the task.
For the purpose of both Experiment 1 and 2, Snoezelen Occupational Therapy treatment was operationally defined as treatment of the participant's proprioceptive and vestibular systems through directed auditory, tactile and visual sensory input to effect arousal change. The participant and the OT alternately adjusted the input through the use of bean bag chair, stationary rocker, projector, light box, ceiling and floor chase lights, fiber optic spray, bubble tube, black lights and new age music synchronized to the light box.
Non-Snoezelen OT treatment is operationally defined as proprioceptive and vestibular calming exercises which preceded the use of rubber stamps and colored pens to make picture communication boards. Therapeutic focus in both situations was to improve visual tracking, attention and midline crossing.
Reliability of the Dependent Measure:
Three Occupational Therapy students, (unfamiliar with the participant and unaware of which type of Occupational Therapy treatment preceded the videotape), independently counted the number of prompts provided during each taped session. It was hoped that by capturing the participant's behavior on videotape immediately after participating in either a Snoezelen session or regular Occupational Therapy session, we could maximize the probability of finding any post session generalization that might occur in his level of engagement in the sandwich making task. By ensuring that the raters were blind to the nature of the preceding type of therapy, we also minimized the probability that their expectations would bias their ratings. Attempts were made throughout the process of reviewing the tapes by the "blind" raters to ensure that their level of agreement remained above 80% for occurrence/non-occurrence agreement, and any observations that did not meet this criteria meant retraining for the observers. This retraining was not necessary during the experiment.
Figure 1 above indicates that during the "A" phase, when Snoezelen was used, the participant went from initially requiring 14 prompts to complete the task to requiring only five prompts three sessions later. During the "B" phase when Snoezelen was not used, he required four prompts initially, then 9 prompts and then 6 prompts. During the final "A" phase when Snoezelen was again used, he required only 3 prompts and then one prompt to complete the task. This met our criteria since the trend during the first "A" phase was downward towards requiring fewer prompts, was upward during the "B" phase, and again downward during the final "A" phase. The trend line for the first "A" phase is drawn through the data to show where the data would be predicted to fall if Snoezelen use were to be continued.
The entire data set, across the three conditions, looks much like a learning curve, in that, over time, there is a decrease in the number of prompts required to complete the task. In order to investigate this explanation of the data, a trend line was drawn through data for the first phase and extended through all three phases. The data demonstrate, weakly, that there seems to be some effect of Snoezelen treatment on the dependent variable, (number of prompts required to complete the sandwich making and eating task). On the other hand, the number of prompts required to complete the task was higher than expected during the first reversal to non-Snoezelen O.T. services during phase 2 and then again lower in phase 3 when Snoezelen use was reinstituted. Weaknesses of the design are the small number of data points collected, and the fact that there was only one reversal. Subsequent reversals, where the data showed the same trend observed during the first three "A" and "B" phases would demonstrate this more effectively. The absence of a baseline prior to beginning Snoezelen treatment also weakens the possibility of demonstrating an effect due to the fact that we cannot observe what would happen to the data in the absence of any intervention. The small number of data points in general, and the fact that this represents data from only one subject, further compromises the researchers' ability to draw conclusions from the experiment. Remember though, that this is only a pilot experiment carried out to test the procedure so it could be successfully replicated with additional subjects over a longer period of time with more frequent reversals.
Participant and Setting:
Participant for Experiment 2 was the same as described above for Experiment 1.
The design for Experiment 2 was the same as that for Experiment 1, except a baseline phase was added with data collected prior to the initiation of any Occupational Therapy sessions. .
Procedure was identical to that used in Experiment 1. Experiment 2 attempts to extend the findings of Experiment 1 by examining whether the change observed in Experiment 1, (more engagement during the post-Snoezelen therapy sandwich making task) is further generalized to the rest of the participant's day and week. The dependent measure here is the frequency of challenging behaviors exhibited by the participant during the next few days until his next Occupational Therapy session. Day Habilitation program staff routinely collect data on the frequency of challenging behaviors exhibited by each consumer on a daily basis. To add rigor to this data collection, once per week, or approximately 20% of the time, a second staff member who was also familiar with the participant was also asked to independently report the frequency of targeted challenging behaviors for the participant in these experiments. This "reliability check" occurred throughout the course of the experiment. Further, day program staff were unaware and therefore blind to which Occupational Therapy method was being used at the time for the participant. For the current participant, four challenging behaviors were being exhibited (elopement, pica, self-injury, crying incidents), but only two, (pica and crying incidents) occurred with enough frequency to be relevant to the research question, (ie., with sufficient frequency to be sensitive to changes over short periods of time). Therefore, data for these two behaviors were collected and graphed in a manner similar to the number of prompts required to complete the sandwich making task.
Behavior: Crying Incidents
For the first challenging behavior, (number of crying incidents per day), the level of inter- observer agreement between staff members counting the number of crying incidents when sampling 20% of occasions was 77.8%, calculating the number of agreements/non-agreements divided by total number of occasions. Based on this information, the mean level during baseline was plotted and this figure extended through subsequent phases of the experiment as a means of assisting visual interpretation of the data. This average during baseline was approximately .5 incidents of crying per day. When we switched to Snoezelen based OT during phase 2, there were 6 sessions in a row of no incidents. There were 9 days below the average for baseline and only 4 above, almost three times as many below as above the mean.
Switching back to non-Snoezelen treatments during the next phase, there were only 6 below and 7 above the mean line for baseline, only slightly better than baseline, with 2 days where the number reaches 2 per day, the maximum observed.
For the final reversal back to Snoezelen based O.T. therapy, there were three days below the mean level for baseline and only one above, again a 3:1 ratio of good to bad days.
Second Behavior: Pica
Data for Pica (number of cigarette butts eaten per day) follows a similar pattern to that of crying incidents, but is even stronger in its support for the difference between Snoezelen and non-Snoezelen phases of the experiment. The level of inter- observer agreement between staff members counting the number of incidents of cigarette butt pica when sampling 20% of occasions was 80%, calculating the number of agreements/non-agreements divided by total number of occasions. Based on this information, during the first Snoezelen phase, there are 9 days below the baseline mean level and only 4 above, again an almost 3:1 ratio. During the next non-Snoezelen phase, there were 11 days above and only 2 below the baseline mean level, a clear change in an undesirable direction, from the previous phase.
During the final reversal back to Snoezelen therapy, there were 3 days below and only one day above the baseline mean level. Again, there was also a difference in the number of days per phase where the maximal number of incidents per phase was reached (2 per day), 2 days during baseline, 1 day during the first Snoezelen phase, 4 days during the non-Snoezelen reversal and none during the final Snoezelen reversal.
There seems to be a fairly clear difference between the level of challenging behaviors exhibited during baseline and non-Snoezelen phases as opposed to the Snoezelen phases for both of the behaviors recorded. This is present in both the ratio of days which were above/below the mean level for baseline as well as the number of days in which the measure reached its maximum value (2 occurrences for both behaviors). The difference is in the direction of fewer challenging behaviors on days following Snoezelen treatment as opposed to days following non-Snoezelen treatments.
The current series of pilot experiments attempted to explore whether behavioral improvements observed anecdotally during Snoezelen Occupational Therapy treatment sessions for a person with moderate mental retardation, autism and severe challenging behaviors generalized beyond the session to two different settings: increases in engagement during tasks immediately following the treatment sessions as well as frequencies of occurrence of two challenging behaviors on the days after treatment sessions.
As discussed above, the experiments provide support to the hypothesis that these effects generalize to both post-session engagement in a task (sandwich making) as well as to the frequency of two challenging behaviors on days subsequent to the OT sessions. As pilot studies, they have several severe limitations. Data is presented for only one subject, making it difficult to generalize these results to others who might benefit from the procedure. Only one treatment reversal is included, making it difficult to determine if other factors might be confounded with the main effect observed, since the strength of the observed connection is increased with increases in the number of reversals which repeat the pattern. The number of data points per phase is also small, increasing the difficulty of observing and extrapolating any trends observed. Finally, there is something curious about the relative strength of the effects observed across the two experiments. The first experiment looks at generalization to a task which follows immediately after the therapy session, while the second experiment looks at generalization across several hours to several days to the frequencies of challenging behaviors, an outcome which is not directly the target of the therapy. One would expect it to be easier to demonstrate generalization across a shorter period of time as well as to a behavior which is more directly targeted by the intervention (task engagement rather than a reduction in the frequency of challenging behaviors). The current experiment found a very weak effect in experiment 1 and a more robust effect in experiment 2. This might be caused by a lack of sensitivity of our engagement measure in experiment 1 as compared to experiment 2, or it may identify a clue about the latency of short-term as opposed to long term effects of therapy.
Subsequent work by our research team will explore possible explanations for this finding in addition to trying to increase its generalizability. The pilot project will be replicated with three additional subjects, using more frequent reversals and phases of longer duration.
Cook, T. D. & Campbell, D. T. (1979). Quasi-experimentation: Design and analysis issues for field settings. Boston, MA: Houghton-Mifflin Co.
Cuvo, A. J., May, M. E., & Post, T. M. (2001). Effects of living room, Snoezelen room and outdoor activities on stereotypical behavior and engagement by adults with profound mental retardation. Research in Developmental Disabilities, 22, 183-204.
Hogg, J., Cavet, J., Lambe, L. & Smeddle, M. (2001). The use of Snoezelen as multi-sensory stimulation with people with intellectual disabilities. Research in Developmental Disabilities, 22, 353-372.
Kazdin, A. E., (1982). Single-case research designs: Methods for clinical and applied settings,. Oxford, UK: Oxford University Press.