mealtime bx article

211

Functional Assessment and Treatment of Mealtime Behavior Problems

Tami L. Galensky St. Louis County Special School District

Raymond G. Miltenberger North Dakota State Univesity

Jason M. Stricker Arlington Development Center

Matthew A. Garlinghouse Southeast Human Service Center

Abstract: This study utilized descriptive assessment methods to develop hypotheses regarding the function of mealtime behavior problems for three typically developing children. Functional treatment was evaluated in the natural setting with caregivers as change agents. Overall, results of the descriptive assessment suggested that each child’s problem behavior was maintained by escape and, to a lesser extent, attention. In addition, this study suggested that direct observa- tion was more reliable than a behavioral interview or questionnaire in acquiring the informa- tion necessary to develop hypotheses on factors maintaining a child’s mealtime behavior problems. Finally, a functional treatment package consisting of extinction, stimulus fading, and reinforcement of appropriate eating behaviors implemented by the caregivers was effective in decreasing the mealtime behavior problems for two of the children who continued in the study, thus providing support for the hypotheses developed from the assessment.

Feeding problems, such as food refusal and food selectiv- ity, are reported to occur frequently in children with men- tal developmental disabilities and/or physical limitations and less frequently in children of typical development (Linscheid, 1983). Jones (1982) reported that 19% to 61% of individuals with mental developmental disabilities dis- played behavioral feeding problems. Although there is lit- tle rigorous experimentation on feeding problems in the general child population (Linscheid, 1983), Sisson and Van Hasselt (1989) reported that feeding problems &dquo;occur in approximately one out of four children&dquo; (p. 45). Babbitt et al. (1994) also indicated that some reviews estimate that feeding disorders may affect 25% of infants and children. The bulk of scientific research has focused on populations composed of children with intellectual, neurological, or physical limitations. Due to the clinical significance of feeding disorders, the research on feeding disorders usually takes place in the inpatient setting and focuses on having the therapist as the change agent.

The literature on food refusal has expanded since the late 1970s and early 1980s. Numerous treatment studies support the use of behavioral interventions, especially posi- tive reinforcement procedures (Ahern, Kerwin, Eicher, Shantz, & Swearingin, 1996; Cooper et al., 1995), stimulus

fading procedures (Freeman & Piazza, 1998; Luiselli, 1994; Shore, Babbitt, Williams, Coe, & Snyder, 1998), and escape extinction procedures (Johnson & Babbitt, 1993; Kern & Marder, 1996) as effective procedures in the treatment of food refusal. Some researchers have used other procedures, such as peer modeling (Greer, Dorow, Williams, McCorkle, & Asnes, 1991 ), caregiver training in behavioral treatments (Werle, Murphy, & Budd, 1993), and relaxation tech- niques, to assist in managing feeding problems (Weinman, Haydan, & Sapan, 1990). However, identifying the vari- ables maintaining food refusal prior to implementing treatment procedures has been rarely reported. In fact, very few experiments have mentioned or included func- tional assessment procedures for food refusal. For exam- ple, although many studies use extinction of refusal to accept food, refusal to swallow food, and inappropriate mealtime behaviors, very few of these studies address or demonstrate the maintaining reinforcers for the behavior prior to treatment implementation.

During the past two decades more emphasis has been placed on the assessment of controlling environmental variables to ascertain the function of problem behavior(s). Functional assessment uses a variety of techniques to iden- tify antecedent and consequent events, including indirect/

212

informant methods, such as interviews, rating scales, and questionnaires, and direct/naturalistic methods, which in- volve direct observation. Functional analysis, the third class of functional assessment techniques, involves the actual manipulation of environmental variables to establish a functional relationship between the controlling variables and the behavior (Lennox & Miltenberger, 1989).

In the literature on food refusal and food selectivity, it appears as though most assessments begin with a clinical interview; however, few of the studies report what sort of interview was used to acquire the reported information. Interestingly, the studies concerned with implementing be- havioral interventions rarely mention the use of a behav- ioral interview and rarely identify the antecedents and consequences related to the problem behavior. It appears as though most researchers informally consider possible hypotheses regarding the function of eating-related prob- lems and very rarely document these hypotheses.

A few studies mentioned the use of behavioral or functional assessment but did not discuss the techniques used or report the results (Hagopian, Farrell, & Amari, 1996; Luiselli, 1989; Riordan, Iwata, Finney, Wohl, & Stan- ley, 1984). A few other researchers have reported descrip- tive functional assessment results prior to treatment for food refusal. Linscheid, Tarnowski, Rasnake, and Brams (1987) noted that a 6-year-old boy with short gut syn- drome needed considerable persuasion from care pro- viders to consume nutritional supplements and that the child engaged in escape behavior. Handen, Mandell, and Russo (1986) used an interview to determine caregiver consequences for food refusal and direct observation to as- sess both antecedents, specifically the use of prompting and consequences. Werle et al. (1993) employed a home- based functional assessment evaluating caregiver feeding behaviors, antecedents such as provision of prompts and food properties, and consequences such as provision of at- tention and rewards. Luiselli (1994) used a behavioral feeding assessment but did not report hypotheses resulting from this assessment. Although the author mentioned (post hoc) that vomiting might have occurred in order to escape the meal, this hypothesis was not assessed. Cooper et al. (1995) used a behavioral interview that looked at an- tecedents and consequences. From this assessment, hy- potheses were developed about the problem behaviors and were utilized in the selection of treatment components. Fi-

nally, Hoch, Babbitt, Coe, Duncan, and Trusty (1995) re- ported a history of escape following food refusal for a 3-year 6-month-old girl but did not mention if this infor- mation was acquired through a prior assessment or by ob- servation during treatment.

To date, a functional analysis of the maintaining vari- ables for food refusal has not been reported in the feeding disorder literature. However, Munk and Repp (1994) eval- uated a behavioral assessment that investigated the effect of food characteristics, such as food type and texture, on

food acceptance in five individuals with disabilities and limited food intake. The experimenters used 10 to 12 dif- ferent food types and 4 types of food texture. The assess- ment consisted of presenting several foods at the same texture level, then using these same foods while progress- ing to coarser textures to determine whether the individual rejected food of a certain type and/or texture or displayed total food refusal. Munk and Repp proposed more research regarding the effects of treatment based on assessment re- sults and suggested functional analysis of consequences.

Although the use of functional assessment and analy- sis procedures is common practice in research that ad- dresses behavior problems in children and persons with developmental disabilities, functional assessment is rarely used as the basis for treatment selection for feeding disor- ders. The purposes of this study were to evaluate indirect versus direct functional assessment methods for develop- ing hypotheses about the function of food refusal and then to evaluate the effects of brief functional treatment con- ducted by the caregivers in the natural setting on the oc- currence of food refusal.

Method

PARTICIPANTS AND SETTING

Participants were three typically developing children who exhibited frequent mealtime behavior problems. Brenda, age 2 years 10 months, engaged in food refusal, elopement, play, and occasional expulsion of nonpreferred food. The caregivers indicated that they had tried various techniques such as verbal prompting, removing Brenda’s plate contin- gent on inappropriate behavior, and providing reinforce- ment with treats and games to get Brenda to eat and behave appropriately during meals.

Roxy, age 6 years 11 months, engaged in food refusal, expulsion of nonpreferred foods, play, and elopement. The caregivers’ attempted techniques included the inconsistent use of punishment, moderate verbal prompting, assisted feeding, and rewards with treats and activities, none of which increased her food intake.

David was a 4-year and 11-month-old boy who dis- played very restricted food preferences and engaged in tantrum behavior if a nonpreferred food was on his plate. During the beginning of this study, David ate only six foods, and his caregivers reported that certain foods that David used to eat were no longer accepted. The caregivers indicated that they occasionally tried to prompt consump- tion of new foods, but to no avail. Therefore, they discon- tinued the use of verbal prompting and began to solely provide preferred foods for meals without requiring David to eat. David’s caregivers reported that the pediatrician in- dicated no physiological abnormalities that would account for his refusal behaviors. All assessments and meals were

213

conducted in the dining areas of the participants’ homes, with the entire family present.

DATA COLLECTION AND DEPENDENT MEASURES

Data were collected for all children on the percentage oc- currence of target behaviors during 10-second intervals, the number of bites per minute, and the estimated per- centage of total food intake, as well as the estimated per- centage of total preferred and nonpreferred food intake. Each session was videotaped and scored according to a 10-second partial interval recording method. A high- resolution VCR that digitally displayed seconds and min- utes was used to view each session. The observer marked each interval in which a target behavior occurred or con- tinued to occur. This procedure allowed for reliable real- time recording of the percentage of intervals the problem behavior occurred and the number of bites of food per minute while eliminating the need for the observer’s pres- ence. Percentages were calculated by dividing the number of intervals the child engaged in the target behavior by the total number of intervals during the meal and multiplying by 100. The number of bites per minute was calculated by dividing the total number of bites by the number of min- utes the meal lasted.

A bite was recorded if the child opened his or her mouth and bit down on the food within 5 seconds after the utensil presentation. Refusal behavior included negative statements indicating that the child did not like the food, did not want to eat, was not hungry, or was feeling sick, and/or physical refusals such as pushing the plate or spoon away or shaking his or her head &dquo;no.&dquo; Refusals were counted as one episode if they occurred together within 5 seconds. Elopement was scored anytime the child’s but- tocks were out of the chair except when he or she was try- ing to reach for something pertaining to the meal (e.g., if a child was reaching for a condiment). An expulsion was scored anytime food, which was previously in the child’s mouth, was spit out beyond the lip or chin area or was taken out of the mouth with his or her fingers. Food play included any manipulation of food with the child’s fingers and any use of a utensil, other than for the goal of picking up the food and directing it to the mouth for ingestion.

In addition to these behavioral measures, the percent- age of each child’s meal consumption and the consump- tion of preferred and nonpreferred foods were recorded by the caregiver. The caregiver completed a rating form every meal. First, the caregiver circled the rating that estimated the child’s total meal intake. Next, the caregiver used the same rating scale to estimate the approximate percentage of preferred and nonpreferred foods consumed. The care- giver used the following rating scale: 1 = 0% of the meal, 2 = 1 % to 24% of the meal, 3 = 25% of the meal, 4 = 26% to 49% of the meal, 5 = 50% of the meal, 6 = 51 % to 74% of

the meal, 7 = 75% of the meal, 8 = 76% to 99% of the mea4 and 9 = 100% of the meal.

Phase 1: Descriptive Assessment

Descriptive assessment involved the completion of a func- tional assessment questionnaire, entitled the Functional Assessment Questionnaire of Mealtime Behavior Problems (FAQ); a functional assessment interview (FAI); and direct observation of the antecedents and consequences pertain- ing to the problem behavior via completion of an A-B-C checklist (Miltenberger, 2001).

FAO .

Questions pertained to the antecedents and consequences of both feeding behavior and mealtime disruptive behav- ior in order to assist in developing hypotheses regarding the function of problem behavior. The questionnaire in- cluded sections on present and past feeding behavior and took approximately 30 to 60 minutes to complete. Interob- server agreement on the hypothesized function of the behavior was assessed by having two behavior analysts in- dependently review the completed questionnaire and record their hypotheses. They also rated their confidence in the hypotheses on a 7-point scale ( 1 = not at all confident, and 7 = completely confident).

FAI

The first author conducted a behavioral interview using the questions from the FAQ. The interview lasted about 60 to 90 minutes, and the caregivers were encouraged to an- swer with as much detail as possible and to provide exam- ples. Following the completion of the interview, the first author used the information to determine the function of the food refusal and then rated the confidence in this hy- pothesis on a 7-point scale. Interobserver agreement on the function of the behavior was gathered by having a practicing master’s-level behavior analyst listen to an audiotape of the interview, record his or her hypotheses re- garding the function of the behavior, and provide a confi- dence rating of the hypotheses.

A-B-C OBSERVATIONS

The A-B-C checklist (Miltenberger, 2001), a form that doc- uments the time the problem behavior occurred and the antecedents and consequences of the behavior, was com- pleted while observing four videotaped meals. Observers were provided operational definitions of the target behav- ior(s) and the possible antecedent and consequent events to record, all of which were acquired from the interview and preliminary mealtime observations. From the A-B-C assessment, the average percentage of times the occurrence

214

of the mealtime problem behavior was followed by the consequences listed on the checklist was determined by di- viding the number of occurrences of each consequence by the frequency of the behavior. A hypothesis about the function of the behavior was identified from the A-B-C

recording by the first and second authors after they inde- pendently evaluated the completed A-B-C data sheets. Both raters also indicated the confidence of their hypothe- ses on a 7-point scale. &dquo;

INTEROBSERVER AGREEMENT OF A-B-C OBSERVATIONS

Interobserver agreement on the scoring of the A-B-C data from videotape was obtained for 50% of the videotaped observations. The observers used the digital second counter on the VCR in order to facilitate reliable recording of the antecedents, consequences, and behavior. The ob- servers recorded the exact time the target behavior occurred and then identified the antecedents and consequences for that behavior. Percentage of agreement was calculated by comparing agreement on the occurrence of the target be- havior at the specific time. Agreements were defined as the two observers identifying the same behavior as occurring within 3 seconds of each other. Percentage agreement on the occurrence of behavior was determined by dividing the number of agreements by the number of agreements plus disagreements and then multiplying by 100. For each in- stance of behavior that was previously agreed on, percent- age of agreement was calculated for both the antecedents and the consequences for the target behavior. Percentage

agreement on the antecedent events was determined by di- viding the number of agreements on antecedents by the number of agreements plus disagreements on antecedents and then multiplying by 100. This same procedure was used to determine percentage of agreement on consequent events.

Agreement between observers on the occurrence of Brenda’s target behaviors ranged from 75% to 100%, with a mean of 87.5%. Interobserver agreement on the occur- rence of antecedents ranged from 97% to 99.3%, with a mean of 98.2%; agreement on the occurrence of conse-

quences ranged from 90.5% to 100%, with a mean of 95.3%. For Roxy, interobserver agreement on the occur- rence of target behaviors ranged from 72% to 81%, with a mean of 76.5%. Agreement between observers on the oc- currence of antecedents and consequences resulted in means of 98.5% (range 98% to 99%) and 98%, respec- tively. For David, agreement between observers on the oc- currence of target behaviors ranged from 81 % to 100%, with a mean of 90.5%. Interobserver agreement on the oc- currence of antecedents and consequences ranged from 89% to 97% and 95% to 98%, respectively, with means of 93% and 96.5%, respectively.

Phase 1: Results and Discussion

PROCEDURE ’

For each participant, the results were examined for the FAI, FAQ, and the A-B-C Checklist. Table 1 reports the hy-

Table 1. Phase 1 Descriptive Assessment Results for Each Child: Hypothesized Functions, Percentage of Interobserver Agreement, and Confidence Ratings

Note. FAI-I = functional assessment interview hypotheses from the primary rater; FAI-2 = functional assessment interview hypotheses from the secondary rater; FAQ-1 = functional assessment questionnaire hypotheses from the primary rater; FAQ-2 = functional assessment questionnaire hypotheses from the secondary rater A-B-C- I= direct observation hypotheses from the primary rater; A-B-C-2 = direct observation hypotheses from the secondary rater. Each rater’s confidence ratings are in parentheses. aoverall mean confidence rating per method. boverall % of agreement on the behavior function between raters.

215

pothesized functions of the mealtime problem behavior for each child, the corresponding confidence ratings from the independent raters, the overall percentage of interob- server agreement between behavior analysts on the func- tion of mealtime problem behavior per assessment method, and the overall mean confidence rating per assessment method.

FUNCTIONAL HYPOTHESES FROM THE FAI

For Brenda, the primary and secondary raters agreed that food refusal behaviors functioned to acquire escape (mean confidence rating of 6.0: moderately certain). The primary rater also hypothesized access to tangibles (confidence rat- ing of 7.0: strongly certain). Attention was also hypothe- sized by the secondary rater (confidence rating of 7.0). For Roxy, both raters supported the escape function (mean confidence rating of 6.5: moderately certain). The primary rater also included attention and access to tangibles as maintaining factors (confidence rating of 4.0 for each function). For David, both the primary and secondary raters determined escape to be a maintaining factor (mean confidence rating of 6.5). In addition, the primary rater hypothesized access to tangibles as a maintaining factor (confidence rating of 7.0).

Overall, across children, the primary and secondary raters’ reviews of the information from the interview re- sulted in a mean confidence rating of 6.1 (moderately cer- tain). In addition, the primary and secondary raters agreed 100% of the time in identifying an escape function and agreed 0% of the time in identifying an attention or tangi- ble function.

FUNCTIONAL HYPOTHESES FROM THE FAO

For Brenda, both the primary and secondary raters agreed that the food refusal behavior functioned to acquire escape (mean confidence rating of 4.5: fairly certain) and atten- tion (mean confidence rating of 4.0). In addition, the primary rater hypothesized that access to tangibles also contributed to maintaining the refusal behavior (mean confidence rating of 3.0: fairly uncertain). For Ro~, the primary and secondary raters indicated that both escape and attention maintained the behavior (mean confidence ratings of 4.7: fairly certain and 2.7: mildly uncertain, re- spectively). In addition, the secondary rater hypothesized that access to tangibles also was a maintaining factor (con- fidence rating of 2.0), thus indicating moderate uncer- tainty. For David, both the primary and secondary raters agreed that the food refusal behaviors functioned to ac- quire escape (mean confidence rating of 4.7: fairly certain). The primary rater also indicated that access to tangibles was a maintaining factor (confidence rating of 4.5), and the secondary rater indicated that attention was a main- taining factor (confidence rating of 4.5).

Overall, across primary and secondary raters, the questionnaire resulted in a mean confidence rating of 4.0 ( fairly uncertain). Across children, the primary and sec- ondary raters agreed 100% of the time in identifying an es- cape function, agreed 67% of the time in identifying an attention function, and agreed 0% of the time in identify- ing a tangible function.

FUNCTIONAL HYPOTHESES FROM A-B-C OBSERVATION

For Brenda, direct observation revealed that her refusal be- haviors acquired escape from eating 70% of the time, ac- quired attention from the caregivers 58% of the time, and gained access to preferred foods 15% of the time. The pri- mary and secondary raters agreed that both escape and at- tention were maintaining factors (mean confidence ratings of 6.3 and 6.5, respectively). For Roxy, the A-B-C assess- ment indicated that mealtime behaviors occurred to es-

cape eating 26% of the time and acquire attention from caregivers 67% of the time. The primary and secondary raters determined that escape and attention were main-

taining factors (mean confidence ratings of 6.8 and 5.5, respectively). For David, refusal behaviors resulted in at- tention 64% of the time, escape 57% of the time, and ac- cess to tangibles 14% of the time. The primary and secondary raters agreed that both escape and attention maintained refusal behaviors (mean confidence ratings of 6.8 and 6.5, respectively).

Overall, the primary and secondary raters’ reviews of the information from direct observation resulted in a mean confidence rating of 6.4 (moderately certain). The primary and secondary raters agreed 100% of the time on the functions of escape and attention and the absence of a

tangible function for all of the children. The results of Phase 1 indicate that direct observation,

via the A-B-C checklist, resulted in hypotheses that yielded the highest confidence ratings, with a mean rating of 6.4, and the highest interobserver agreement, with a mean per- centage agreement of 100%. Therefore, the hypotheses from direct observation were used in the development of functional treatment.

The FAQ resulted in hypotheses of attention and es- cape that were similar to the hypotheses developed from direct observation; however, it also resulted in the lowest confidence ratings from independent raters, with a mean rating of 4.0, and low interobserver agreement, with a mean percentage agreement of 56%. Although the ques- tionnaire results led to the hypotheses of escape and atten- tion, the information acquired from the caregivers was sparse and lacking detail, thus resulting in the lower confi- dence ratings.

On the other hand, the interview yielded higher confi- dence ratings from independent raters, with a mean rating of 6.1, but the lowest interobserver agreement, with a mean percentage of 44%. Although the interview took much

216

time to conduct, it provided more detailed information and practical examples of feeding behaviors and the events surrounding feeding situations than did the questionnaire.

In light of these results, some issues need to be con- sidered. First, no decision rules are in place to assist indi- viduals in determining the function of a behavior. This issue is especially true with interpreting A-B-C assessment results and may have influenced the results from this

study’s A-B-C assessment. For example, the A-B-C assess- ment indicated that David’s food refusal behaviors resulted in escape 57% of the time, attention 64% of the time, and access to preferred food 14% of the time. Therefore, rele- vant questions may have included, (a) Was the primary hypothesis an attention or escape function? (b) How significant was the function of gaining access to preferred foods or other tangibles? (c) Should intervention have addressed all of the demonstrated hypotheses? and (d) What percentage of time should a consequence follow a behavior before it is considered a significant function of the behavior?

Another factor requiring consideration involves the influence of differing levels of experience of the indepen- dent observers on the determination of hypotheses from the functional assessment methods. For example, a behav- ior analyst with a doctorate and more experience and fa- miliarity with the procedures may be more efficient and accurate in determining hypotheses as compared to a less experienced, master’s-level behavior analyst. A third issue involves the factors influencing the independent raters’ in- terpretation of the interview and questionnaire results. For example, the first author conducted all of the interviews and viewed and recorded all of the A-B-C data, whereas the independent raters reviewed the permanent products (e.g., the completed questionnaire, the audiotaped interview, and the scored A-B-C checklist). Therefore, the personal interaction with the family or the procedures may have re- sulted in differing hypotheses than the hypotheses devel- oped from those individuals viewing the raw data alone.

Overall, this study found that neither the results from the questionnaire nor the behavioral interview were as consistent as the results from direct observation in de-

termining functional hypotheses. Until further research provides more support for the utility of interview or ques- tionnaire methods, these results suggest that researchers/ practitioners should use direct rather than indirect func- tional assessment methods when attempting to develop hypotheses about the function of mealtime problem be- haviors. Although the current feeding literature briefly dis- cusses the use of interviews and direct observation, it is important that future studies systematically evaluate func- tional assessment and functional analysis procedures with feeding behaviors to ascertain variables that influence and maintain these behaviors. In addition, one direction for fu- ture research may be to compare the results of functional

assessment procedures carried out by caregivers versus re- searchers.

The direct observation results from this phase of the study, as well as results from the preference assessment in Phase 2, were used to develop functional interventions to address mealtime problem behaviors (Phase 3).

Phase 2: Stimulus Preference and Avoidance Assessment

PROCEDURE

Twelve foods, determined via caregiver report, were evalu- ated in a paired-choice format (Fisher et al., 1992) to de- termine relative stimulus preference and avoidance (Fisher et al., 1994). First, the caregivers of each child developed a list of six preferred foods and six nonpreferred foods that the caregivers wanted the child to be eating by the termi- nation of treatment. During the assessment, each food item was placed in a clear plastic cup with a plastic spoon to facilitate consumption. Two cups were then placed on the table approximately 5 inches from the child and ap- proximately 4 inches apart. The child was instructed to choose only one food item from the pair. Preference was determined by measuring the number of times a child chose an item out of the total number of times that item was presented.

PREFERRED VERSUS NONPREFERRED FOODS

We briefly compared preferred versus nonpreferred foods to assess the validity of the caregiver’s verbal report re- garding preferred and nonpreferred food items. Each of the six preferred foods was randomly paired with one of the six nonpreferred foods and presented to the child in six trials. Validity of the caregiver’s verbal report was evaluated by comparing the assessment results (the actual food items that the child chose) with the caregiver’s predictions. To assess predictions, the caregiver reviewed the list of the preferred/nonpreferred food pairs and circled the food item he or she thought the child would choose.

STIMULUS PREFERENCE ASSESSMENT

The stimulus preference assessment compared preferred food items in order to determine potential edible rein- forcers for use during treatment. Each of the six preferred food items was randomly presented once with every other preferred food item. In this forced-choice assessment, each food item was presented 5 times, for an overall total of 15 trials. Relative preference was reported as the number of times each food item was chosen across the 5 trials that in- cluded that food.

217

STIMULUS AVOIDANCE ASSESSMENT

The stimulus avoidance assessment compared each of the six nonpreferred food items in order to develop a hierarchy of the most-acceptable to least-acceptable nonpreferred food items. Each nonpreferred food was randomly pre- sented once with every other nonpreferred food. Each nonpreferred food was presented 5 times, for a total of 15 trials. Relative preference of nonpreferred foods was re- ported as the number of times each food item was chosen across the 5 trials that included that food.

Phase 2: Results and Discussion

PREFERRED VERSUS NONPREFERRED FOOD

All three children chose the preferred foods over the non- preferred foods.

STIMULUS PREFERENCE ASSESSMENT

Brenda’s preference assessment indicated that macaroni was the most preferred food because it was chosen five out of five stimulus presentations. The remaining order of pre- ferred foods consisted of cheese (4/5), cereal (3/5), cooked carrots (2/5), and french fries (1/5). Roxy’s preference as- sessment indicated that ice cream was the most preferred item because it was chosen five out of five presentations, followed by cereal (3/5), grapes (3/5), bread (2/5), waffles (1/5), and noodles (1/5). David’s preference assessment in- dicated that chicken nuggets, chosen four out of five pre- sentations, was the most preferred stimulus, followed by hash browns (3/5), slightly toasted bread (3/5), waffles (3/5), hamburgers (2/5), and apples (0/5).

STIMULUS AVOIDANCE ASSESSMENT

The results of the stimulus avoidance assessment (see Fig- ure 1) indicated that, for Brenda, raw carrots, chosen five out of five presentations, was the most acceptable nonpre- ferred food item, followed by mashed potatoes (4/5), boiled potatoes (3/5), rice (1/5), chunked beef (1/5), and crumbled beef (1/5). For Roxy, the most acceptable non- preferred food was watermelon (5/5), followed by can- taloupe (4/5), mashed potatoes (3/5), green beans (2/5), peas (1/5), and corn (0/5). David’s results indicated that bread with peanut butter (5/5) would be the most accept- able nonpreferred food, followed by mashed potatoes (4/5), grapes (3/5), corn (2/5), cheese pizza (1/5), and mac- aroni (0/5). In this assessment, David refused to take a bite of any nonpreferred foods; therefore, the most-acceptable to least-acceptable nonpreferred food range was assessed by having him indicate which food item was more accept- able. In addition, five random stimulus pairs of food items

Figure 1. The number of nonpreferred food presen- tations accepted for each child during the stimulus avoidance assessment.

were represented to assess the reliability of his verbal re- port. Results indicated that the reliability of his verbal re- port was 100%.

INTEROBSERVER AGREEMENT AND CAREGIVER’S

VERBAL REPORT

An independent data collector scored the videotaped pref- erence assessment and stimulus avoidance assessment for each participant. Interobserver agreement was calculated by dividing the number of times there was agreement on the child’s food choice by the number of agreements plus disagreements and then multiplying by 100. The mean in- terobserver agreement for each child’s food choice was 100%.

For all of the children, each caregiver was 100% cor- rect in identifying preferred over nonpreferred foods. For Brenda, the caregiver was 93.3% accurate in predicting the child’s preferred food choice and 26.7% accurate in pre- dicting the child’s nonpreferred food choice, with a total accuracy of 60%. For Roxy, the caregiver was 53.3% accu- rate in predicting the child’s preferred food choice and

218

60% accurate in predicting the child’s nonpreferred food choice, with a total accuracy of 56.7%. For David, the care- giver was 73% accurate in predicting the child’s preferred food choice and 83% accurate in predicting the child’s nonpreferred food choice, with total accuracy of 78%.

Phase 2 results suggested that the caregiver’s verbal re- port was valid in that each caregiver correctly reported the child’s preferred and nonpreferred foods and correctly pre- dicted the child’s consumption of the preferred food over the nonpreferred food. Altogether, the caregivers were not much more accurate than chance in predicting the child’s food choice in both the stimulus preference and avoidance evaluations.

Overall, the stimulus preference and avoidance assess- ments were useful in systematically answering important questions relevant to developing functional treatment. First, the stimulus preference results assisted in the deter- mination of the most preferred food to use as a potential reinforcer for each child. Second, the stimulus avoidance assessment results assisted in the determination of the least

nonpreferred food to use as the initial target stimulus in order to increase the child’s success with a stimulus fading procedure.

The literature currently supports the use of preference assessments in identifying reinforcers (Bowman, Piazza, Fisher, Hagopian, & Kogan, 1997) for use during treat- ment. One important issue to consider is that an individ- ual’s relative preference for a stimulus may change from day to day (Bowman et al., 1997). Unfortunately, we were not able to determine the degree to which this phenome- non may have influenced the results of this assessment. Be- cause an individual’s preference may change periodically, it seems relevant to suggest that a preference assessment may need to be conducted more than once, especially when re- lying on a stimulus to act as a reinforcer for appropriate behavior. Therefore, further research needs to be done to determine if more frequent preference assessments are es- sential, and if so, can we determine more efficient ways to conduct these assessments on a frequent basis? Due to the small sample size, replication of this phase is encouraged to determine if the results of caregiver verbal report occur across individuals and to provide further evidence for or against determining relative nonpreference.

Phase 3: Treatment Evaluation

PROCEDURE

Phase 3 involved using the information acquired from Phase 1 for determining the components of a functional treatment package and using the information from Phase 2 for determining potential positive reinforcers as well as a hierarchy of target foods to be used in treatment. A multi- ple baseline across participants’ research design was used

to evaluate brief functional treatment implemented by the caregivers in the natural setting.

Baseline Measures

Meals were videotaped in the dining area of the partici- pants’ homes with the experimenter present but not inter- acting with the child or caregiver. For Brenda, Roxy, and David, baseline consisted of 6, 9, and 11 videotaped din- ners, respectively. For each participant, the caregiver pre- sented at least one nonpreferred food with the meal and proceeded to conduct the mealtime as usual. In addition, the caregiver occasionally verbally prompted the child to try the nonpreferred food item.

Functional Treatment .

For each child, functional treatment consisted of rein- forcement and extinction procedures that specifically ad- dressed the function of each child’s mealtime behavior. Functional treatment was implemented by the caregivers in the natural setting with the assistance of the therapist, who provided the caregivers with guidance and corrective feedback. The treatment components were the same across children because the direct observation assessment indi- cated escape and attention functions for all three children. Functional treatment consisted of the following:

1. tangible reinforcement, in which preferred foods were given contingent on bites of nonpreferred food;

2. stimulus fading, in which the portion size of nonpreferred food was increased across meals; treatment sessions progressed from the least

.

, nonpreferred to the most nonpreferred foods (as determined in the stimulus avoidance assess- ment). Components 1 and 2 were designed to

’

. address the escape function of the problem behaviors;

3. differential reinforcement of alternative/other behavior, in which attention was contingent on the appropriate use of utensils, sitting in the chair, eating, and abstaining from play behaviors every 30 seconds during a meal in order to address the attention function of the problem behaviors;

4. extinction, which involved ignoring attention- maintained problem behaviors (to address the attention function), and escape extinction, in which the child was not allowed to leave the table until the nonpreferred food was eaten (to address the escape function);

5. physical redirection, involving gentle redirection of any occurrence of elopement behavior and the second occurrence of play behavior (to address the escape function).

219

Interobserver Agreement

For each participant, two independent data collectors scored 25% of the videotaped sessions from both baseline and functional treatment phases. Interobserver agreement was calculated by dividing the number of intervals of agreement on the occurrence and nonoccurrence of the target behavior by the total number of intervals in the sample and then multiplying by 100. An agreement was scored when the observers marked the exact same behav- ior in the exact same interval. The mean interobserver

agreement for Brenda’s refusals was 95.1% (range = 89%-98.6%), play was 86.5% (range = 76.4%-100%), elopement was 97.7% (range = 95.7%-100%), and expul- sions was 98.5% (range = 96.9%-100%). The mean agree- ment for the occurrence and nonoccurrence of bites was 94.6% (range = 91.7%-97%). For Roxy, mean interob- server agreement was 98.4% for refusals (range = 95.8%- 100%), 98% for elopement (range = 93%-100%), 99.2% for expulsions (range = 98%-100%), 93.5% for play (range = 84%-98.9%), and 94.3% for the occurrence and nonoccurrence of bites (range = 90%-99%). The mean in- terobserver agreement for David was 93.6% for refusals

(range = 85.3%-100%), 99.9% for expulsions (range = 99.5%-100%), and 89.7% for the occurrence and nonoc- currence of bites (range = 70%-100%).

Treatment Integrity

Twenty-five percent of each participant’s baseline and treat- ment sessions were scored for the occurrence of the fol-

lowing treatment components:

(a) contingent attention and preferred food for con- sumption of the targeted nonpreferred food;

(b) ignoring of refusal and play behaviors; (c) physical redirection for elopement and play

behaviors; and (d) differential reinforcement, an average of every

30 seconds, for appropriate eating behavior, sitting in the chair, and the absence of play.

Contingent reinforcement of the acceptance of the tar- geted nonpreferred food item was determined if praise and a bite of preferred food were delivered following the con- sumption of the target food. Ignoring of refusal and play behaviors was determined by recording whether attention followed either behavior, thus indicating an incorrect in- tervention response on the caregiver’s behalf. The correct use of physical redirection was determined if assistance was applied following the occurrence of elopement and/or play behaviors. Differential reinforcement (DRA-30s) was calculated by dividing the number of 10 second intervals in which behavior-specific praise occurred by the total number of intervals.

Social Validation

Treatment acceptability was assessed via the Treatment Evaluation Inventory-Short Form (TEI-SF; Kelly, Heffer, Gresham, & Elliott, 1989), a questionnaire with nine items rated on a Likert-type scale from 1 to 5. A score of 27 or greater, out of a total possible score of 45, on the TEI-SF indicated that the child’s caregiver perceived the treatment package to be acceptable for treating the child’s mealtime problem behaviors.

Phase 3: Results and Discussion

FUNCTIONAL TREATMENT

Figure 2 displays the percentage of occurrence of play and elopement behaviors for Brenda and Roxy, and Figure 3 displays the percentage of refusal and expulsions for each meal across baseline and treatment. For Brenda, treatment resulted in a mean decrease in refusal, elopement, and play behaviors. Expulsions were low across baseline and treat- ment. For Roxy, treatment resulted in a decrease of elope- ment and play behavior, whereas expulsions and refusals remained low across baseline and treatment phases. For David, treatment resulted in a &dquo;burst&dquo; of refusal behavior,

Figure 2. The percentage of play and elopement be- haviors for Brenda and Roxy across baseline and treatment sessions.

220

Figure 3. The percentage of refusal behavior and ex- pulsions for Brenda and Roxy across baseline and treatment sessions.

which was followed by his parents’ decision to terminate treatment. Bites per minute increased for both Brenda and

Roxy (see Figure 4). Figure 5 shows an increase in the per- centage of nonpreferred foods consumed by Brenda and Roxy.

Brenda

For Brenda, food refusal decreased from a baseline mean of 16.4% to a treatment mean of 7.9% of intervals. Play be- haviors decreased from 25.6% to 10.6% of intervals. For

expulsions, baseline and treatment means were low (1.3% and 1.8%, respectively). Elopement remained stable, with a baseline mean of 10.7% and a treatment mean of 10.1 % of intervals, respectively. The number of bites per minute slightly increased from a baseline mean of 2.2 to a treat- ment mean of 2.6. The percentage of preferred foods con- sumed slightly increased from baseline to treatment, with mean rankings of 5.5 (50%-75% of food consumed) and 6.8 (75% of food consumed), respectively. The percentage of nonpreferred foods consumed increased greatly from

baseline, with a mean ranking of 1.5 (0%-25% consump- tion), to treatment, with a mean ranking of 9.0 (100% con- sumption).

ROXV

For Roxy, play behaviors decreased from a baseline mean of 17.9% to a treatment mean of 6.4% of intervals. Elope- ment decreased from 14.7% to 1.2% of intervals. Expul- sions decreased slightly, from 2.4% to 0.8% of intervals. The percentage intervals of food refusal remained stable, with a mean of 3.4% in baseline and 3.6% in treatment. The number of bites consumed per minute increased from 2.0 in baseline to 3.0 in treatment. The percentage of pre- ferred foods consumed remained stable during baseline and treatment sessions, with mean rankings of 7.9 and 7.5 (consumption of 75%-100% of food), respectively. The percentage of nonpreferred foods consumed increased from baseline, with a mean ranking of 2.6 (0%-75% of foods consumed per meal), to treatment, with a mean ranking of 9.0 ( 100% consumption).

Figure 4. The number of bites per minute for Brenda and Roxy across baseline and treatment sessions.

221

Figure 5. The caregivers’ rating of the percentage of preferred and nonpreferred food consumption per meal for Brenda and Proxy across baseline and treatment sessions.

David

For David, food refusal increased from a baseline mean of 22.9% to a treatment mean of 79.8% of intervals before functional treatment was terminated by his parents. Expul- sions were very low across baseline and treatment ( 1.1% & 0.6% of intervals, respectively). The percentage of preferred foods that was consumed slightly increased from baseline to treatment, with mean rankings of 6.7 (consumption of 50%-75%), and 8.5 (consumption of 75%-100%), respec- tively. The percentage of nonpreferred foods that was con- sumed increased from baseline, with a mean ranking of

1.4 (consumption of 0%-25% of food), to treatment, with a mean ranking of 9.0 ( 100% consumption).

. TREATMENT INTEGRITY

For Brenda, data from the baseline videotapes showed that the caregiver provided behavior-specific praise in 1.8% of the 10-second intervals. During treatment, differential re- inforcement for appropriate mealtime behavior was pro- vided 26% of the 10-second intervals; physical redirection for elopement was provided 100% of the time it was needed; and contingent preferred food and praise for eat- ing the nonpreferred target food was provided 100% of the time, thus indicating the correct use of these treatment procedures. On the other hand, physical redirection for play behaviors was provided approximately 50% of the time; planned ignoring of play behavior occurred approx- imately 50% of the time; and attention was provided for refusals 27% of the time; thus suggesting lower integrity with these treatment components.

For Roxy, praise was provided in 0% of the 10-second intervals during baseline. During treatment, contingent preferred food and praise for the consumption of the non- preferred target food was provided 100% of the time; physical redirection for elopement was provided 100% of the time; and differential reinforcement for appropriate eating behavior was provided in 24% of the 10-second in- tervals, thus indicating the correct use of these treatment components. On the other hand, physical redirection for play behaviors was provided approximately 50% of the time; planned ignoring of play behaviors occurred approx- imately 50% of the time; and attention was provided for refusal behaviors 33% of the time, thus suggesting lower integrity for these treatment components.

David’s baseline data indicated that praise was pro- vided in 1.4% of the 10-second intervals. During treat- ment, ignoring food refusal occurred 82% of the time. The use of behavior-specific praise and contingent preferred food following consumption of the target food was used correctly. In addition, the escape extinction procedure and physical assistance was used 100% correctly.

SOCIAL VALIDATION

The caregivers’ total TEI-SF ratings of treatment for Brenda and Roxy were 41 and 32, respectively, thus suggesting that the caregivers found treatment to be acceptable for their children’s behavior problems. Because David did not con- tinue the study, treatment acceptability ratings were not available.

Phase 3 evaluated functional treatment packages de- signed to address each child’s mealtime behaviors, which were hypothesized from the descriptive assessment to be maintained by escape and attention. The findings from

222

Phase 3 suggested that a functional treatment package com- posed of stimulus fading, reinforcement, and extinction decreased the occurrence of food refusal, elopement, and play behaviors for Brenda and decreased elopement and play for Roxy. Treatment also increased the number of bites per minute and increased the percentage of nonpre- ferred food consumed for these two participants. In addi- tion, the extinction burst from David suggested that we were correct in identifying the maintaining variables for his behaviors. However, treatment could not be evaluated

fully for David because the parents terminated their par- ticipation in the study.

Most of the food refusal literature focuses on dealing with feeding problems on the inpatient unit with the ther- apist as the change agent. Research has indicated that treat- ment in a setting other than the natural setting may lead to difficulties in generalizing the results (Bakken, Milten- berger, & Schauss, 1993). A few studies in the food refusal literature have attempted to programfor generalization of treatment results from inpatient therapists to parents as therapists and/or from an inpatient setting to the natural setting (Bernal, 1972; Greer et al., 1991; Handen et al., 1986; Hatcher, 1979; Ives, Harris, & Wolchik, 1978; Thompson, Palmer, & Linscheid, 1977), with only one study to date formally evaluating assessment and treatment in the nat- ural setting (Werle et al., 1993). Similar to Werle et al., we assessed the child in the home and had the caregivers im- plement the treatment package in the home with the fam- ily present; however, certain problems were evident.

First, although functional treatment was effective in improving mealtime behavior, treatment effects were some- what variable. The problem behaviors that occurred at a low level during baseline (expel for Brenda; expel and re- fusal for Roxy) did not decrease during treatment, perhaps because of a floor effect. On the other hand, behaviors that occurred at a moderate to high level in baseline all de- creased at least to some extent during treatment. Moderate treatment integrity may help explain the variable treat- ment results. During baseline for all participants, little praise was provided for eating behavior and attention, and escape usually followed inappropriate behaviors. However, following intervention, the caregivers maintained above 80% treatment integrity with the provision of praise, approxi- mately 100% in the use of contingent preferred food and physical redirection of elopement behaviors, and only ap- proximately 50% integrity with providing physical redirec- tion for play behaviors and ignoring refusal and expulsion behaviors. Overall, a moderate level of treatment integrity was maintained when we were present to provide feedback and guidance. Needless to say, there could have been inade- quate treatment integrity in our absence, thus influencing the consistency of the treatment effect on the behavior.

Poor integrity with the intervention for play behaviors may have resulted from the relative level of importance of the behavior to the caregivers. For example, two of the

caregivers indicated that the play behaviors were not as im- portant as the child eating. This may be why the caregivers were not as diligent with the interventions designed for play behaviors. Second, the topography of play behavior was variable and difficult to observe and record. For exam-

ple, play behaviors of the two participants were occasion- ally subtle, which was demonstrated by the child sucking on the spoon after taking a bite of food. Not only were there problems with the integrity of treatment with play behavior but as mentioned earlier, there was also a lower interobserver agreement on the occurrence of play behav- ior. Behaviors such as play also influenced the determina- tion of the onset and offset of bites, thus leading to lower agreement on this behavior as well. Scoring the occurrence of bites was typically more accurate when the caregiver de- livered each bite of food.

A second problem evident when conducting treat- ment in the natural setting was that it was more difficult to control extraneous factors, such as the participant’s inter- action with a sibling, when the entire family was present. In addition, providing immediate feedback on his or her performance during a family meal was difficult and often discouraged by the caregiver. Therefore, we relied on writ- ten and verbal instruction prior to and following the ses- sion, as well as modeling during the session, in order to train the caregivers in the correct use of the procedures.

A third problem with conducting treatment in the natural setting dealt with the occurrence of more severe re- fusal behavior due to the stimulus control of the caregivers and others present, as well as stimulus control of the set-

ting itself. The possibility of stimulus control was an im- portant consideration for David’s results. Implementation of functional treatment for David resulted in an extinction burst. This burst of refusal behavior involved high levels of crying, screaming, asking to leave, and squirming in his seat in an attempt to leave the table, which occurred for ap- proximately 90 minutes during both treatment sessions. Intervention for this behavior required the use of a physi- cal assistance procedure to keep David in his seat, thus not allowing him to escape the meal. In addition, the &dquo;non- removal of the spoon&dquo; (escape extinction) procedure was used to eliminate the avoidance of swallowing a bite of nonpreferred food. This procedure involved holding an ex- tremely small portion (the size of a pencil eraser) of the most acceptable nonpreferred food (stimulus fading) to David’s lips until acceptance of the bite. Although the care- givers reported that they were prepared for the burst, in re- fusal behavior, it seems they did not comprehend the potential severity of the behavior because when they were faced with dealing with this intense behavior, they opted to terminate treatment. Therefore, we were not able to evalu- ate the results of functional treatment for this participant.

In hindsight, we have to ask what we could have done differently to deal with these events. Functional treatment could have been partially evaluated by excluding the escape

223

extinction procedure and by relying on the stimulus fading procedure with the contingent attention and preferred foods. However, David’s behaviors frequently resulted in escape; thus these procedures alone were not likely to be effective. Lastly, inpatient treatment may have been recom- mended in order to eliminate the effects of stimulus con- trol of the environment and the caregiver and to be better equipped to deal with the extinction burst. One question raised by this study is, &dquo;Did David’s mealtime behavior warrant the necessity of an intensive intervention for food refusal?&dquo; This question broaches some of the theoretical is- sues resulting from this study. Two out of the three fami- lies in this study believed that treatment was acceptable and that the results were beneficial. However, for the third

family, it is not clear in retrospect that David’s behavior problems and lack of intake and variety were severe enough to warrant the more restrictive intervention.

General Discussion

This study addressed some of the gaps in the food re- fusal literature by evaluating functional assessment and preference assessment methodologies with inappropriate mealtime behaviors, by examining the results of brief functional treatment on three typical children’s mealtime behaviors in the natural setting with the caregivers as the change agents, and by gathering treatment integrity and treatment acceptability data. In this investigation, treat- ment improved some target behaviors while leaving others unaffected. Eating behaviors improved for all children (more nonpreferred foods consumed, more bites per minute), and some problem behaviors decreased for Roxy and Brenda, who both completed the study. A floor effect may have been responsible for the lack of improvement in the other target behaviors. That is, the behaviors that were

already low in baseline were the ones that did not decrease further. The behaviors that were occurring at a moderate to high level all decreased with treatment.

This study has produced a number of important find- ings :

1. Direct observation of food refusal behaviors resulted in functional hypotheses that yielded higher confidence ratings and interobserver agreement than interview and questionnaire methods.

2. A stimulus preference assessment was able to identify preferred foods that could function as reinforcers.

3. A stimulus avoidance assessment was helpful in developing a hierarchy of the level of acceptabil- ity of the nonpreferred foods that were to be used as the target stimuli for treatment.

4. A functional treatment package implemented by the caregivers, with moderate treatment

integrity, in the natural setting was effective in decreasing the occurrence of play and elopement behaviors, increasing the number of bites per minute and increasing the percentage of non- preferred food consumed for the two partici- pants who continued in the study.

5. The treatment acceptability results suggested that the caregivers found the treatment package to be acceptable for the two participants who continued in the study.

In addition, the results of treatment provided support for the hypotheses developed from the descriptive assessment. However, because a functional analysis was not conducted, the results do not rule out other potential maintaining factors.

Further research on the function of inappropriate feeding behavior is warranted in order to reduce the nega- tive influence on the family during meals and to facilitate the learning of appropriate and adaptive eating behavior. Emphasis should be placed on conducting rigorous exper- iments to ascertain the various factors influencing and maintaining these behaviors, as well as conducting com- ponent analyses to evaluate individual treatment compo- nents. Likewise, more dismantling and constructive treatment evaluations should be conducted to determine essential treatment components. Results from rigorous treatment evaluations, functional assessments, and func- tional analyses can facilitate the determination of neces- sary treatment components, hence increasing treatment efficiency and decreasing unnecessary treatment duration.

ABOUT THE AUTHORS

Tami L. Galensky, MS, is an applied behavior analysis coor- dinator for the St. Louis County Special School District in St. Louis, MO. Her interests include teaching language and other functional behaviors to children with autism. Raymond G. Miltenberger, PhD, is a professor of psychology at North Dakota State University with research interests in the func- tional assessment and treatment of problem behaviors. Jason M. Stricker, MS, is a behavior analyst at Arlington Develop- mental Center in Arlington, TN. His research interests are in the analysis and treatment of difficult behaviors of individu- als with developmental disabilities. Matthew A. Garling- house, MS, is a psychologist at Southeast Human Service Center in Fargo, ND. He has interests in behavioral interven- tions for persons with developmental disabilities. Address: Raymond G. Miltenberger, Department of Psychology, North Dakota State University, Fargo, ND 58105. e-mail: ray- miltenberger@ndsu. nodak. edu

AUTHORS’ NOTE

This study was conducted as a thesis by the first author. We thank Rebecca Johnson, John Rapp, and Vicki Lumley for their assistance in data collection.

224

REFERENCES

Ahern, W. H., Kerwin, M. E., Eicher, P. S., Shantz, J., & Swearingin, W. (1996). An alternating treatments comparison of two intensive interventions for food refusal. Journal of Applied Behavior Analysis, 29, 321-332.

Babbitt, R. L., Hoch, T. A., Coe, D. A., Cataldo, M. E., Kelly, K. J., Stackhouse, C., & Perman, J. A. (1994). Behavioral assessment and treatment of pedi- atric feeding disorders. Journal of Developmental and Behavioral Pediatrics, 15, 278-291.

Bakken, J., Miltenberger, R. G., & Schauss, S. (1993). Teaching parents with mental retardation: Knowledge versus skills. American Journal on Mental Retardation, 97, 405-417.

Bernal, M. E. (1972). Behavioral treatment of a child’s eating problem. Jour- nal of Behaviour Therapy and Experimental Psychiatry, 3, 43-50.

Bowman, L. G., Piazza, C. C., Fisher, W. W., Hagopian, L. P., & Kogan, J. S. (1997). Assessment of preference for varied versus constant reinforcers. Journal of Applied Behavior Analysis, 30, 451-458.

Cooper, L. J., Wacker, D. P., McComas, J. J., Brown, K., Peck, S. M., Richman, D., Drew, J., Frischmeyer, P., & Millard, T. (1995). Use of component analy- ses to identify active variables in treatment packages for children with feeding disorders. Journal of Applied Behavior Analysis, 28, 139-153.

Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P., Owens, J. C., & Slevin, I. (1992). A comparison of two approaches for identifying rein- forcers for persons with severe and profound disabilities. Journal of Ap- plied Behavior Analysis, 25, 491-498.

Fisher, W., Piazza, C. C., Bowman, L. G., Kurtz, P. F., Sherer, M. R., & Lach- man, S. R. (1994). A preliminary evaluation of empirically derived conse- quences for the treatment of pica. Journal of Applied Behavior Analysis, 27, 447-457.

Freeman, K. A., & Piazza, C. C. (1998). Combining stimulus fading, rein- forcement, and extinction to treat food refusal. Journal of Applied Behav- ior Analysis, 31, 691-694.

Greer, R. D., Dorow, L., Williams, G., McCorkle, N., & Asnes, R. (1991). Peer- mediated procedures to induce swallowing and food acceptance in young children. Journal of Applied Behavior Analysis, 24, 783-790.

Hagopian, L. P., Farrell, D. A., & Amari, A. (1996). Treating total liquid refusal with backward chaining and fading. Journal of Applied Behavior Analysis, 29, 573-575.

Handen, B., Mandell, F., & Russo, D. (1986). Feeding induction in children who refuse to eat. American Journal of Diseases in Children, 140, 52-54.

Hatcher, R. P. (1979). Treatment of food refusal in a two-year-old child. Jour- nal of Behaviour Therapy and Experimental Psychiatry, 10, 363-367.

Hoch, T. A., Babbitt, R. L., Coe, D. A., Ducan, A., & Trusty, E. M. (1995). A swallow induction avoidance procedure to establish eating. Journal of Be- havior Therapy and Experimental Psychiatry, 26, 41-50.

Ives, C. C., Harris, S. L., & Wolchik, S .A. (1978). Food refusal in an autistic type child treated by a multi-component forced feeding procedure. Jour- nal of Behavior Therapy and Experimental Psychiatry, 9, 61-64.

Johnson, C. R., & Babbitt, R. L. (1993). Antecedent manipulation in the treat- ment of primary solid food refusal. Behavior Modification, 17, 510-521.

Jones, T. W. (1982). Treatment of behavior-related eating problems in re- tarded students: A review of the literature. In J. H. Hollis & C. E. Meyers (Eds.), Life threatening behavior: Analysis and intervention (pp. 3-26). Washington, DC: American Association on Mental Deficiency.

Kelley, M. L., Heffer, R. W., Gresham, F. M., & Elliott, S. N. (1989). Develop- ment of a modified treatment evaluation inventory. Journal of Psy- chopathology and Behavioral Assessment, 11, 235-247.

Kern, L., & Marder, T. J. (1996). A comparison of simultaneous and delayed reinforcement as treatments for food selectivity. Journal of Applied Behav- ior Analysis, 29, 243-246.

Lennox, D. B., & Miltenberger, R. G. (1989). Conducting a functional assess- ment of problem behavior in applied settings. Journal of the Association for Persons with Severe Handicaps, 14, 304-311.

Linscheid, T. R. (1983). Eating problems in children. In C. E. Walker & M. C. Roberts (Eds.), Handbook of clinical child psychology (pp. 616-639). New York: Wiley.

Linscheid, T. R., Tarnowski, K. J., Rasnake, L. K., & Brams, J. S. (1987). Be- havioral treatment of food refusal in a child with short-gut syndrome. Journal of Pediatric Psychology, 12, 451-459.

Luiselli, J. K. (1989). Behavioral assessment and treatment of pediatric feed- ing disorders in developmental disabilities. In M. Hersen, R. M. Eisler, & P. M. Miller (Eds.), Progress in behavior modification (Vol. 24, pp. 91-133). Newbury Park, CA: Sage.

Luiselli, J. K. (1994). Oral feeding treatment of children with chronic food re- fusal and multiple developmental disabilities. American Journal on Mental Retardation, 98, 646-655.

Miltenberger, R. G. (2001). Behavior modification: Principles and procedures. Pacific Grove, CA: Wadsworth.

Munk, D. D., & Repp, A. C. (1994). Behavioral assessment of feeding prob- lems of individuals with severe disabilities. Journal of Applied Behavior Analysis, 27, 241-250.

Riordan, M. M., Iwata, B. A., Finney, J. W., Wohl, M. K., & Stanley, A. E. (1984). Behavioral assessment and treatment of chronic food refusal in

handicapped children. Journal of Applied Behavior Analysis, 17, 327-341. Shore, B. A., Babbitt, R. L., Williams, K. E., Coe, D. A., & Snyder, A. (1998).

Use of texture fading in the treatment of food selectivity. Journal of Applied Behavior Analysis, 31, 621-633.

Sisson, L. A., & Van Hasselt, V. B. (1989). Feeding disorders. In J. K. Luiselli (Ed.), Behavioral medicine and developmental disabilities (pp. 45-73). New York: Springer Verlag.

Thompson, R. J., Palmer, S., & Linscheid, T. R. (1977). Single-subject design and interaction analysis in the behavioral treatment of a child with a feed- ing problem. Child Psychiatry and Human Development, 8, 43-53.

Weinman, B., Haydan, S., & Sapan, J. (1990). Behavioral treatment of a food refusal conversion disorder in a mentally retarded adult. Journal of Mental Deficiency Research, 34, 501-508.

Werle, M. A., Murphy, T. B., & Budd, K. S. (1993). Treating chronic food re- fusal in young children: Home based caregiver training. Journal of Applied Behavior Analysis, 26, 421-433.

Action Editor: Robert L. Koegel