Expressive Language As Predictor for Amount of Assessment Required to Evoke Challenging Behavior

Background:  

When behavior analysts are enlisted to reduce challenging behavior, valid assessments are needed to develop individualized treatment. The functional analysis (FA) is the most valid measure for this purpose; it methodically tests and confirms hypothesized functions of challenging behavior. The Latency-Based Functional Analysis (LBFA) is an expedited FA model validated for confirming evocative and reinforcing variables of challenging behavior.  Length of the LBFA is predicated on the amount of time required to systematically evoke and reinforce challenging behavior, then replicated with experimental control.

Within a randomized controlled trial (RCT) assessing for potential cost-benefit of behavior analytic services within typical inpatient hospital settings, we conducted latency-based FAs of the challenging behavior of [SJ1] 18 children diagnosed with autism and with varying communicative abilities.

Objectives:  

Test the hypothesis that patient expressive language skill may predict the speed with which challenging behavior can be evoked within LBFAs.

Methods:  

Eighteen children (17 male, 1 female) ages 6-16 (mean 10.3) with autism participated in this study.  Twelve were admitted to a university-based child and adolescent psychiatric hospital and six were admitted to a medical floor of a university-based children’s hospital. All had history of chronic challenging behavior.  We conducted all assessments in inpatient units of these settings; either in participants’ rooms or multi-purpose rooms. 

Definitions for each participant’s target behaviors were operationalized based on caregiver report during pre-FA interviews. LBFAs were conducted using procedures based on Thomasson-Sassi, et al. (2011). Minimum numbers of assessment series required to evoke target behaviors for each participant were recorded. A series was defined as the total set of 2-4 assessment session types (e.g. attention, play, tangible, escape) to be presented within the LBFA. Patients’ expressive language skills were rated based upon direct observations by behavior analysts during patient interviews, assessments, and informal observations within the hospital setting. Patient utterances were sorted into two categories: “limited vocal” (patients used approximations of sentences or shorter phrases with limited vocabulary) or “fully vocal” (speaking primarily in full sentences and using expansive vocabulary).

Finally, we evaluated the degree to which expressive language rating predicted the number of series needed to evoke challenging behavior within the LBFA using a point-biserial correlation.

Results:  

Results indicated expressive language rating was positively correlated with the number of series needed to evoke challenging behavior (r_pb = .58). Higher verbal ability predicted a higher number of series.

Conclusions:  

Results suggest patients demonstrating high levels of expressive language skill may require adaptation of LBFAs for improved results. Limitations of this study include small sample size, and pilot iteration of a language skills assessment. Because language performance was rated based on post-hospitalization reviews of videos capturing participant interactions within behavioral assessments, the experimenters did not have the opportunity to conduct a validated language assessment with the participants. Future research should use previously-validated language measures, further test the expressive language skills rating scale used within this study, verify which adaptations to behavioral assessments can be useful to rapidly evoke and study challenging behavior for verbally skilled patients, and examine replicability.