Steps to Effective Problem-solving in Group Homes

Abstract

Aggressive/challenging behaviors (A/CB) are a major public health problem for individuals with intellectual disabilities (ID). A leading reason for psychiatric hospitalizations and incarcerations, such behaviors are costly to the health care system, agencies, and families. Social problem-solving (SPS) training programs for individuals with ID have had positive behavioral results, but most were conducted in clinical or forensic settings. None was a community-based preventive intervention, none examined whether the behaviors decreased in residential and work settings, and none addressed cost-effectiveness. In preliminary work, we modified an effective SPS training program (ADAPT: Attitude, Define, Alternatives, Predict, and Try out), using input from individuals with ID and residential staff, as a community-based preventive intervention that we delivered in group homes (STEPS: Steps to Effective Problem-solving). Individuals with ID have high rates of obesity, and our attention-control condition is a nutrition intervention: Food for Life. We describe the protocol for a randomized clinical trial to: (1) test the efficacy of the STEPS intervention for improving SPS skills and reducing A/CB compared to an attention-control nutrition intervention in group homes; (2) assess the mediating effect of residential staff SPS skills, group-home level SPS skills, and group cohesiveness on the improvement of SPS skills and reductions in A/CB; and (3) evaluate the cost-effectiveness of STEPS. We expect to show that STEPS is a preventive strategy to reduce A/CBs among individuals with ID and improve the cost-effectiveness of their care.

1. Introduction

Aggressive/challenging behaviors (A/CBs), including destruction of property and threat of or real personal injury, are a major public health issue for individuals with intellectual disabilities (ID) and their support systems. Such behaviors among individuals with ID are a leading reason for emergency department visits [1], psychiatric hospitalizations [2], and incarcerations [3]. A single trip to the ED may cost $1,500 or more [4].

More than 530,000 individuals with ID in the U.S. now live in nonfamily residential facilities, with 77% residing in small group homes (typically 4–6 residents) [5]. In group homes, residential staff provide assistance with residents’ many needs. The small group home environment can decrease the social distance between residential staff and residents and encourage social networks among individuals with ID [6]. If provided with appropriate assistance, the group home has potential to be an ideal environment for individuals with ID to obtain A/CB support. Fewer A/CB problems are reported in group homes than in larger facilities [7], but individuals with ID living in group homes have higher rates of A/CBs than those who live with their families [8–10].

Intellectual disability is characterized by deficits in social problem-solving (SPS), [11,12]. the cognitive and behavioral activities (attitude and style) one uses to recognize, cope with, and find solutions to problems. Individuals with ID who have A/CBs tend to have a negative attitude and to view interpersonal situations as hostile [12]. Individuals with ID who do not use A/CBs use more assertive responses (rational style) [13]. Individuals with ID who have A/CBs respond to situations with hostile actions more frequently than non-aggressive individuals with ID, and, in stressful situations, use more aggressive responses (impulsive style) [13].

SPS training interventions have shown some success in reducing A/CBs among individuals with ID, but were conducted in clinical treatment and forensic settings [14–20]. Our modification, Steps to Effective Problem-Solving (STEPS), is a community preventive intervention. It is a 6-session, 12-week program with one booster session at 18 weeks that uses the group home support system, including residential staff [21]. STEPS is based on the ADAPT model (Attitude, Define, Alternatives, Predict, and Tryout)[12]. We pilot tested STEPS in two group homes. Findings from our pilot indicated that individuals with ID could provide examples of problems likely to lead to A/CBs and, with prompting, identify immediate emotional responses likely to trigger A/CBs. Among individuals with ID, the intervention improved SPS skills and decreased A/CBs with effect sizes of d = .6 for each. Residential staff also improved SPS skills with an effect size of d = .6. Findings indicated that SPS interventions can decrease A/CBs and have effects on residential staff skills [21,22].

In this paper, we describe the research protocol for our ongoing clinical trial of STEPS, including development of the intervention, how we identify homes that meet criteria, how we identify individuals with ID and residential staff in the homes who meet criteria, our randomization procedure to the intervention and attention-control nutrition program, and the training and fidelity plans for the clinical trial.

2. STEPS Conceptual Framework

The STEPS Framework (Figure 1) was used in our preliminary work and is now used in our clinical trial. The STEPS Framework is based on the Interaction Model of Client Health Behavior [23,24] and the Relational/Social Problem-Solving Model [12] which are both grounded in the broad philosophic construct of human agency, which addresses the capacity of humans to adapt, change, make choices, and make things happen by their own actions [25]. Elements of the STEPS Framework include baseline determinants of the A/CBs of individuals with ID, intervention strategies, the support environment for SPS, and subsequent outcomes (SPS skills and behaviors). The framework specifies that participant outcomes (i.e., SPS and A/CBs) are dynamically related: the greater the improvement of SPS skills, the greater the likelihood of decreased A/CBs.

Figure 1:

Steps to Effective Problem-Solving (STEPS) Framework

2.1. Determinants of A/CBs

Background characteristics include demographics, past life events, environment, current health, including depressive symptoms and medication use, and baseline SPS skills.

2.2. Intervention strategies

SPS is made up of two independent but interrelated dimensions: attitude (positive or negative) and style (rational, avoidant, or impulsive) [12]. The intervention’s strategies are targeted to SPS skills, specifically to increase positive attitude and rational SPS style [12]. The dimension of attitude includes positive and negative attitudes. Positive attitude involves recognizing problems and their sources and believing in one’s ability to manage or solve problems. Negative attitude involves thinking of problems as a threat, inaccurately describing their sources, and believing that one is unable to solve or manage the problems [12]. The three SPS styles are rational, avoidant, and impulsive. Defining problems, generating and thinking through alternatives, and systematically carrying out and verifying solutions are part of the rational problem-solving style [12]. Inaction, dependence, and passivity toward problems are part of the avoidant style. Immediate emotional responses to problems are part of the incomplete, hurried, and careless impulsive style of SPS. Aggressive/challenging behaviors are associated with negative attitude and impulsive SPS styles [13].

The STEPS intervention affects relationship between the baseline background characteristics and the outcomes of SPS skills and A/CBs. Targeting SPS skills for improvement can reduce A/CB outcomes. Aggressive/challenging behaviors were measured through a coded videotape of a group problem-solving interaction, scores on the General Maladaptive Index, and incident reports.

2.3. Support environment for SPS.

• 2.3.1SPS of residential staff

  The results of two previous studies addressing A/CBs suggested that outcomes were better for individuals who had a staff member from a program setting accompany them, but staff were not included as an integral part of the interventions [26,27], so the relationship between residential staff SPS skills and the SPS skills and A/CBs of individuals with ID residing in the home is not known. Our study systematically involves residential staff and addresses the mediating effect of residential staff SPS on the SPS skills and A/CBs of individuals with ID living in the home.

• 2.3.2.SPS of group

  Group training in SPS skills may encourage individuals with ID to think about the point of view of others and identify alternatives for problem solutions [19]. Previous research showed that when individuals with mild or moderate ID living in residential facilities made decisions as a group about common problems, the decision-making skills of the individuals improved [28]. We measure group SPS from coding a videotape of a group problem-solving interaction. Our study systematically addresses the mediating effect of group SPS skills on the SPS skills and A/CBs of the individuals with ID residing in the homes.

• 2.3.3Factors in the group environment, such as group cohesiveness, have been shown to affect outcomes of research [29].

3. Developing the STEPS Intervention

Developing the STEPS intervention took place in two phases: tailoring ADAPT concepts for individuals with ID and their residential staff and piloting the tailored program.

3.1. Tailoring ADAPT model

In our work to develop the STEPS program, we used multiple sequential methods and sought input from four groups of stakeholders to modify and tailor the ADAPT model of SPS as a community preventive intervention for individuals with ID living in group homes. In work previous to developing the STEPS program, we found the views of individuals with ID on their mood and experiences differed from those of their support staff from various settings and caregivers [30–32]. Individuals with ID answer questions about their experiences of depression than staff answer questions about them [30=32]. Also, translating clinical research to community settings included obtaining input from community members on how they understood materials and how they would like interventions delivered [33]. The multi-step process included that: (1) initial modifications to the ADAPT program were made and the initial STEPS manual was developed, and presented to an advisory committee of 6 residential agency program directors (from 4 different agencies) who were responsible for developing behavioral programs for individuals with ID. Revisions were made based on their feedback. (2) We conducted cognitive interviews with three individuals with ID who lived in group homes and who had a history of A/CBs. Cognitive interviews are a way to get information on how populations understand, process, and respond to information, and on what might not be understandable, and where there might be breakdown in the delivery of a program [34, 35]. Our cognitive interviews with individuals with ID provided examples of problems likely to lead to A/CBs and ways the individuals might respond, alternative wording, and ideas for the logistics of the program. Based on the cognitive interviews, further revisions were made to the STEPS manual. (3) Cognitive interviews were conducted with three group home residential staff. (4) After all of the cognitive interviews were completed, an expert panel and a methodological consultant were convened to edit and approve the program used during the pilot study. (5) After the pilot was concluded, we conducted follow-up satisfaction interviews with residents and residential staff who participated in the pilot and further modified the STEPS program. (6) The same three individuals with ID who participated in the cognitive interviews during development of the manual then reviewed the modifications and gave approval to the program [21]. The intervention was then offered in two homes with high rates of A/CBs to residents and residential staff as a group, with the expectation that the same staff would participate regularly. Each session was approximately one hour in length [21,22].

3.2. STEPS pilot research

Following development of the STEPS manual, a pilot test of the STEPS intervention was conducted using a pre-post design [22]. Approval for the study was given by the IRB at Rush University Medical Center.

• 3.2.1Recruitment

  For the pilot, we first recruited two group homes (one male, one female) with high rates of A/CBs. The homes had to have at least an average of one incident report per month over the last three months, and at least 30% of the residents engaged in A/CBs that warranted an incident report.

• 3.2.2Participants

  After recruiting homes, we recruited staff in the home and individuals with ID in the homes. We recruited four staff from the homes (two from each home) and 12 adults with ID, five in a female home and seven in a male home, (out of 14 residents in the two homes).

• 3.2.3Measures

  In the pilot, we used the same measures now being used in the clinical trial (Table 1). The measurement tools measure the elements of the STEPS Framework. It should be noted that the Iowa Family Interactions Rating Scales (IFIRS) [36,37] are used to measure four constructs: (1) individual SPS skills, (2) individual A/CBs, (3) residential staff SPS skills, and (4) group SPS skills. Thus, rating of the IFIRS is an important aspect of our measures. For the IFIRS, a videotape of a problem-solving interaction of the group (individuals with ID and residential staff) is made in each group home. The data collector has the group choose a problem to discuss, including discussing what they think the problem is, what happens, when it happens, who is involved, what they want to happen, what they think are solutions, and what they think are the best solutions. The discussion lasts about 7–11 minutes, and no more than 11 minutes are coded.

  The videos are sent to Iowa State University for rating. Rating procedures for the videos include detailed directions on what interaction behaviors to assess, rating frequencies, intensity, affective tone, context of behaviors, and rating the proportions of interactions in which behaviors are displayed. Each item is scored on a 9-point scale (1 = not at all characteristic to 9 = mainly characteristic). Raters receive intensive and ongoing training. Interrater reliability is measured on 5% of the videotapes. If there is greater than one-point difference between scores on an item, the raters meet to determine a consensus score [36,37]. The IFIRS is a well-established measure that has predictive validity across basic and applied research [36,37]. Research indicates three factors of underlying types of communication in the interactions: negativity (such as anger and hostility), positivity (such as warmth and closeness), and problem-solving effectiveness (evidence of working toward a solution) [38]. Although observational ratings are costly, because raters are naïve to condition, the IFIRS has the advantage of providing objective ratings of SPS and A/CBs versus subjective self-report or caregiver report. Though raters are naïve to study condition when coding videotapes, it is possible that something within the video may reveal condition. This is expected to be rare and should have minimal impact on the behavioral ratings.

  • 3.2.3.1
    Individuals with ID outcomes
    SPS skills are assessed in two ways.

    • 3.2.3.1.1
      The IFIRS Individual-level Problem-solving Scales [36,37] are used to objectively assess solution quantity, solution quality, effective process, disruptive process, and negotiation/compromise at the individual level from the videotapes (see 3.2.3 for rating of the IFIRS).
    • 3.2.3.1.2
      The Problem-Solving Task (PST) [15] is used to objectively measure the problem-solving skills of individuals with ID. This measure has four questions on each of five problem vignettes that measure SPS skills. The vignettes are read by our data collectors to the individuals, and responses are audiotaped. In our preliminary work, graduate nursing students were successfully trained to rate the PST.
  • 3.2.3.2
    Individuals with ID outcomes
    Aggressive/challenging behaviors are measured in three ways.

    • 3.2.3.2.1
      The IFIRS Dyadic-Interaction Scales [36,37] are used to measure hostility, verbal attack, physical attack, contempt, etc. (total 22 items), of each individual with ID. The same procedures are used as above with the IFIRS and scored by staff at Iowa State University.
    • 3.2.3.2.2
      General Maladaptive Index (GMI; subjective) is part of the Inventory for Client and Agency Planning (ICAP), an instrument that agencies in this study must use by State regulation [39]. The GMI was used because it is already part of agency records and is a well-established measure which has good convergent validity with the Behavior Problems Inventory, which in turn has good convergent validity with the Aberrant Behavior Checklist, Diagnostic Assessment for the Severely Handicapped-II, and Nisonger Child Behavior Rating Form [40,41]. Qualified Intellectual Disability Professionals at the residential agency and the work setting supervisor (sheltered workshops and employment) are asked to fill out the GMI on participants with ID at the intervals required by our study. The GMI measures the frequency (0 = never to 5 = one or more times an hour) and severity (0 = not serious to 4 = extremely serious) of problem behaviors in eight domains (hurtful to self or others, destructive, disruptive, socially offensive, unusual/repetitive, withdrawn/inattentive, uncooperative). The GMI has a mean of 0, SD of 10, and scores above −11 are considered “normal” [39]. Reported test/retest reliability is 0.80, and inter-rater reliability is 0.80 [39]. For the purposes of this research, the GMI is scored using ICAP CompuScore software with an algorithm that compares participants’ scores to standardized scores in different age groups.
    • 3.2.3.2.3
      Incident reports are the third measure (objective) of A/CBs. Incident reports are obtained from agency records. Agency incident reports have common elements, including checklists of key aspects of the incidents (type of problem, who was involved, location of incident, interventions, if third party involvement [police, paramedics], and outcomes [injury, property damage, hospitalization]). Body figures to mark locations of any injuries are included. Residential staff members fill out incident reports. The reports are reviewed by Qualified Intellectual Disability Professionals, kept in resident files, and tracked by the agencies. Data are abstracted from agency tracking records and de-identified. We analyze counts of incidents for outcomes and summarize descriptive data of key elements of the incident reports for the 24 weeks prior to baseline and Weeks 12, 24, and 36 (Week 36 is 24 weeks after the intervention).
  • 3.2.3.3
    Aggressive and challenging behavior determinants

    • 3.2.3.3.1
      Demographics (age, gender, ethnicity, and level of ID) are obtained from agency records. Epidemiologic studies show that, among individuals with ID, A/CBs are more of a problem among males [42].
    • 3.2.3.3.2
      Life events Life events are measured by the Life Events section of the Psychiatric Assessment Schedule for Adults with Developmental Disabilities. Qualified intellectual disability professional staff at the agency respond to this measure [43]. Among individuals with ID, it is predictive of emergency department visits for A/CBs, of psychopathology, and of psychiatric events [44–46].
    • 3.2.3.3.3
      Environment includes whether homes are urban/suburban and the number of people living in the homes.
    • 3.2.3.3.3
      Current health. Current health includes depressive symptoms of individuals with ID and medication management. Depressive symptoms. The Glasgow Depression Scale for People with Learning Disabilities [GDS-LD]) [47] was developed for use among individuals with ID (called “learning disabilities” in the UK). Items are scored on a 3-point Likert scale, with potential scores ranging from 0–40. In screening for depression, 13 is a useful cutoff score for referral, and, at that score, the GDS-LD has sensitivity ranging 90–96% and specificity ranging 83.9–90% [30,47]. Aggressive/challenging behaviors may be an atypical feature of depression in this population [21]. Rates of depressive symptoms are as high as 28%−39% in individuals with ID [30, 48–50], and more common among females [51]. Medication management. Information on psychotropic medications is gathered from agency records.
  • 3.2.3.4
    Support environment for SPS
    This includes the residential staff SPS, group SPS, and group cohesiveness.

    • 3.2.3.4.1
      Residential staff SPS
      The residential staff SPS is measured in two ways. The IFIRS individual-level problem-solving scales [36,37] are the same instrument used with individuals with ID, and are also score for residential staff. Also, a self-report problem-solving measure, the Social Problem-Solving Inventory Revised – Short Form (SPSI-R SF), is used [52]. The five dimensions of this measure are positive attitude, negative attitude, rational style, impulsive/careless style, and avoidant style.
    • 3.2.3.4.2
      Group SPS is measured with IFIRS group-level problem-solving scales [36,37]. The scales include problem-solving enjoyment, agreement on problem description/solution, implementation, and problem difficulty for group, and are scored by staff at Iowa State University.
    • 3.2.3.4.3
      Group cohesiveness is measured using the cohesiveness subscale of the Group Environment for the Intervention Scale (GEIS) [29]. The GEIS is a 25-item measure of group environments. The three subscales are implementation and preparedness, counterproductive activity, and cohesiveness [29]. In previous research, GEIS scores related to health behavior outcomes [29]. In our study, The GEIS is scored by trained research assistants using audiotapes from Sessions 1 and 6. In our pilot study, graduate nursing students were trained to do GEIS rating, and we are doing the same in the clinical trial [22].
• 3.2.4Methods

  A psychiatric nurse conducted the intervention with a research assistant who was present to help the interventionist during intervention sessions. There were six weekly sessions and one booster session at 12 weeks. All of the sessions lasted about 60 minutes and were conducted in the evenings with the group of adults with ID and the residential staff in their group homes.

• 3.2.5Results

  The mean age of individuals with ID in the pilot study was 36.6 (SD 10.5). The four residential staff and 5 of the 12 individuals with ID were women. Two residential staff (50%) and three individuals with ID (25%) were minorities. The staff and residents were expected to attend all 7 sessions. One individual with ID dropped out over the 12 weeks due to lack of interest, one residential staff member moved to a different home, and another staff member left the employ of the agency midway through the study. Participants with ID attended 70% of the sessions, and residential staff attended 67%. Both the participants with ID (91%) and residential staff (87%) reported being highly satisfied. The effect size for improvement in social problem skills using the IFIRS was d = .60 for individuals with ID and residential staff. The effect size for decrease in A/CBs exhibited by individuals with ID using the IFIRS was d = .60. Effect sizes for improvement were d = .51 in group-level problem-solving (based on IFIRS scores) and d = 1.43 in cohesiveness (Sessions 1–6) using the cohesiveness subscale of the GEIS [29]. According to residential staff, the training in breaking down problems and getting to know how individuals with ID related to each other were the most helpful aspects of the intervention [22].

• 3.2.5Implications

  The results of the pilot study indicated that it was feasible to implement STEPS in a clinical trial within a small community group home with high levels of satisfaction for both adults with ID and residential staff. Based on our work, we finalized the STEPS Framework for our clinical trial (see Figure 1 and explanation in Section 2).

Table 1.

Concepts/Measures, Reliability/Validity, Participant Burden, Source, and Data Collection Time Point for Behaviors

Concept/Measure	Reliability/Validity	Range	Mins	Participants	Sourcea	Schedule (weeks)
Individuals with ID (IWID) outcomes						0	12	24	36
SPS skills  o IFIRS Individual-level Problem-solving scales (5 items coded from videotaped interactions43,44	Predictive, convergent, and discriminant validity	5–45	30	IWID	V	X	X	X
o Problem-solving Task (PST) (5 problems, 20 items, 0–5 scale coded from audiotape)18	Alpha .88–93 .83 interrater .79 test-retest	0–100	20–30	IWID	A	X	X	X
Behaviors  o IFIRS Dyadic-interaction scales (22 items coded from videotaped interactions)43,44	Predictive, convergent, and discriminant validity	22–198	N/A	IWID	V	X	X	X
o GMI scale of ICAP GMI) (8 domains, 0–5 frequency, 0–4 severity)46	.80 interrater .80 test-retest	0–40 0–32	N/A	IWID	R, W	X	X	X	X
o Behavior incident reports (collected weekly)	N/A		N/A	IWID	R	X	X	X	X
A/CB behavioral determinants						0	12	24	36
• Demographics: Age, gender, ethnicity, level of ID	N/A		N/A	IWID	R, S	X
• Life events: Life events section of Psychiatric Assessment Schedule for Adults with Developmental Disabilities (17 items 2-recent 1-event 0-never)50–53	Convergent validity IWID	0–34	15–20	IWID	R	X	X	X
•Environment: Agency characteristics (urban/suburban, # of clients, types of services, # of homes), Home characteristics (agency, location, gender, # of residents)	N/A		N/A		R	X
• Current health  o Glasgow Depression Scale -LD (20 items, 0–2)37,54	Alpha .87–.90	0–40	15–20	IWID	S	X	X	X
o Medication management	N/A		N/A	IWID	R	X	X	X
Support environment for SPS						0	12	24	36
• Residential staff SPS  o SPSI-R SF (25 items, 0–5 scales)59	Alpha .88–.93	0–125	15–20	Res staff	S	X	X	x
o IFIRS Individual-level Problem-solving scales (5 items coded from videotaped interactions)43,44	Predictive, convergent, and discriminant validity	5–45	N/A	Res staff	V	X	X	X
• Group SPS  o IFIRS Group-level Problem-solving scales (4 items coded from videotaped interactions)43,44	Predictive, convergent, and discriminant validity73	4–36	N/A	Group	V	X	X	X
• Group cohesiveness  o Group Intervention Environment Scale (Sessions 1 and 6)36	Alpha .8750	8–32	N/A	Group	A	X	X

^aSource- V for video, A for Audiotape, S for self, R for Agency record, W for work setting record

4. STEPS Clinical Trial Research

4.1. Specific aims and hypotheses

The specific aims of the STEPS clinical trial are listed below.

• 4.1.1Aim 1
  Assess the efficacy of the STEPS intervention in group homes to improve social problem-solving skills and reduce A/CBs of individuals with ID compared to the attention-control nutrition intervention: Food for Life. We address this aim by comparing longitudinal data collected four times over 36 weeks (at baseline, 12, 24, and 36 weeks).

  • We hypothesize that individuals with ID in the STEPS condition will have: (a) improved social problem-solving skills at 12 and 24 weeks and (b) decreased A/CBs in both group homes and in their work settings (e.g., work centers, shelter workshops, supported employment, and independent community employment) compared to the attention-control condition at 12, 24, and 36 weeks, controlling for behavioral determinants of A/CBs.
• 4.1.2Aim 2
  Assess the mediating effect of the support environment for SPS (residential staff SPS skills, group home level SPS skills, and group cohesiveness) on the improvement of SPS skills and reductions in A/CBs of the individuals with ID.

  • We hypothesize that the support environment for SPS will mediate intervention effects on SPS skills and A/CBs in individuals with ID.
• 4.1.3Aim 3
  Evaluate cost effectiveness of STEPS relative to usual care for A/CB incidents in group homes.

  • We hypothesize that STEPS will be cost-effective, taking into account costs of A/CBs borne by the agencies, health care system, public services, and participant and family, compared to usual care (control group costs of A/CBs excluding the costs of the nutrition intervention).

4.2. Methods

• 4.2.1What we incorporated from pilot

  Based on residential staff recommendations and review by residential staff involved in Phases 1 and 2 and our field notes, we developed an orientation manual for residential staff. Based on our experience in the pilot, orientation for residential staff in the clinical trial is conducted before the start of the intervention, and we guide residential staff throughout the program. Also, we now space out the six content sessions over 12 weeks (1 every other week) to allow time to assimilate and practice problem-solving skills, and we have one booster session six weeks later (midway through the maintenance phase, at Week 18). Residential staff suggested that issues such as conflict with relatives are more sensitive, so sessions should start with “easier” issues (e.g., who will sit in the front passenger seat of the van on the way to work) and move to increasingly difficult topics. They also suggested that the research team should ask them about which topics are more sensitive for the residents in the home. For sustainability, and with consultation from our advisory group and others, we also determined to use interventionists with the qualifications of Qualified Intellectual Disability Professionals, not psychiatric nurses. Qualified Intellectual Disability Professionals typically provide service coordination and case management for residents with ID, and are required to have a bachelor’s degree in a human service field with at least one year of experience working with individuals with ID [53]. Going into the clinical trial, this meant further scripting of the interventions.

• 4.2.2Overview and study design

  The STEPS project uses a two-group intervention, attention-control clinical trial. One group receives the STEPS intervention and the other the Food for Life nutrition intervention. A cluster-randomized clinical trial design is being used, in which 36 group homes (18 male and 18 female) are randomly assigned to either the STEPS intervention (n = 18; 9 male, 9 female homes) or the attention-control Food for Life condition (n = 18; 9 male, 9 female homes). We expect to complete 9 rounds of the interventions with four homes in each round. For each round, we plan to have 2 female homes (1 in the STEPS intervention, 1 in the Food for Life intervention) and 2 male homes (1 in the STEPS intervention, 1 in the Food for Life intervention).

• 4.2.3Power analysis

  Statistical power was estimated using Optimal Design for Longitudinal and Multilevel Research software [54]. In our preliminary study, an effect size of d = .60 was obtained for improvement in SPS scores of individuals with ID. In a community-based SPS intervention study with a wait-list control group among individuals with personality disorders (and no ID), an effect size of d = .56 [55] was found (similar to the effect size of d = .6 for changes in SPS in our pilot study). Our baseline data suggests that SPS and A/CBs had site intraclass correlations (ICCs) close to .00. Considering that effect sizes might have been smaller in our pilot data if we had a control group, for the clinical trial, we use the more conservative effect size of d = .56 [55] for changes in problem-solving. Also, due to the small sample size in our pilot study, an ICC of .02 may be too low; thus, we use a more conservative estimated ICC of .06. With this effect size (d = .56) and ICC (.06), an average cluster size of four residents, assuming α = .05 and 32 sites, the power of the clinical trial is .81. This represents a sample of 144 Individuals with ID. Based on previous research, we conservatively estimate 20% attrition [15,55,56]. Thus, we expect to recruit 180 individuals with ID (an average of 5 per group home × 36 homes) and 36–108 residential staff (1–3 per group home).

• 4.2.4Participants

  The study participants are adults (over age 18) with mild to moderate ID living in group homes located in a major metropolitan area in the United States and the residential staff who work with them. Individuals with mild to moderate ID have an IQ of 50–75 [57] and basic reading and writing skills ranging from being able to write their names and addresses to writing simple sentences; hold jobs in work centers, supported employment, or independent community employment; may take public transportation independently; and need support in instrumental activities of daily living such as budgeting, making appointments, and nutrition planning. The homes that participate in the STEPS study are gender-specific, and each home generally has two or more residential staff working in the evening. Most of the small community group homes were formerly single-family homes and look like other homes in the neighborhoods.

• 4.2.5
  Inclusion/exclusion criteria

  • 4.2.5.1
    Group home criteria
    The inclusion criteria for the group homes are that they have (1) individuals with mild to moderate ID; (2) at least 10 A/CB incident reports during the prior six-months; (3) at least 30% of participants with A/CB incidents during the prior six months;(4) five or more residents; and (5) a minimum of three residents and one residential staff who agree to participate. Group homes are excluded if they exclusively serve individuals with ID who also have serious mental health issues (e.g., schizophrenia or bipolar disorder), homes exclusively for people with ID who are on a forensic program, and homes exclusively for people with ID and autism. (We do not exclude individuals with serious mental illness, who are in forensic programs, and/or who have autism if they live in homes that serve individuals with ID in general).
  • 4.2.5.2
    Individuals with ID criteria
    The inclusion criteria for the individuals with ID include: (1) mild to moderate ID (operationalized as IQ 50–75 [57] per agency records and mild to moderate limitations in adaptive functioning [measured by the Inventory for Client and Agency Planning [39]]); (2) aged 18 or older; and (3) able to speak English and communicate verbally.
  • 4.2.5.3
    Residential staff criteria
    The inclusion criteria for residential staff include: (1) employed as residential staff in the chosen group homes; (2) aged 18 or older; and (3) able to speak English and communicate verbally.

4.3. Recruitment strategy

We have three steps in recruitment: recruitment of agencies, recruitment of homes, and recruitment of staff and residents in the home. We have two randomization steps, the first being agencies and the second of homes within agencies (See Figure 2 [CONSORT])

Figure 2.

CONSORT Flow Diagram for STEPS

• 4.3.1Recruitment and randomization of agencies

  We initially approached 11 agencies and received letters of support from all of them. We randomly ordered the 11 agencies for participation. However, some of the agencies that initially agreed to participate in this study have undergone changes that make participation difficult, including state budget cuts that meant that they closed some group homes, residential staff shortages, and merging with other agencies. We were unable to recruit homes at all of the initial 11 agencies. Keeping the initial order for the remaining randomized agencies, the research team are approaching additional local agencies to discuss the research and are randomizing those agencies that agree. We are recruiting until we have a total of 36 homes: 18 in the STEPS condition (9 male and 9 female) and 18 in the Food for Life condition (9 male and 9 female).

• 4.3.2.Recruitment and randomization of group homes

  Once agencies agree to be sites for the research, agency administrators identify homes that meet our criteria. We next randomize homes that meet criteria. We hope to get at least a cohort of four homes (two of each gender) that meet criteria. Two homes are assigned to STEPS (1 male, 1 female) and 2 to Food for Life (1 male, 1 female). In the event that an agency has fewer than four eligible homes (two for each gender), we randomize the eligible homes with homes from another agency where four homes are also unavailable. If necessary, we adjust ordering of agencies to allow all participating homes within an agency to participate within the same cohort. Group homes (residents and residential staff) are recruited after randomization to simplify the process of explaining the study to individuals with ID.

• 4.3.3Recruitment of staff in the group homes

  Once homes are randomized, residential staff are recruited first because there must be at least one residential staff person enrolled from the home to include the residents with ID. Agency administrators first identify staff and then distribute a recruitment flyer and the consent forms for review. Once consent to contact is completed, a research team member meets the staff member to explain the study, confirm eligibility, answer questions, and complete the consent process. If no residential staff are willing to consent to participation, that group home is ineligible and another group home serving the same gender is approached for recruitment.

• 4.3.4Recruitment of individuals with ID in the group homes

  Once at least one staff member from the home agrees to participate, agency administrators obtain consent to contact from the guardians or individuals who are their own guardian. Then a research team member contacts guardians and individuals with ID who are their own guardians to explain the study and answer questions. If the person is interested, a consent form is given to them. In some cases, the agency administrators prefer to obtain consent and then send the completed consents to the research team. Individuals with ID who are not their own guardians are asked to assent for the study. For individuals with ID, regardless of whether they have a guardian, a research team member uses an assessment informed by Fisher and Cea’s [58] to assess ability of individuals with ID to consent/assent to participate in the research. Individuals with ID who are unable to consent/assent are not accepted. If fewer than three residents have consent for participation, that group home is ineligible and another group home serving the same gender is approached for recruitment.

4.4. STEPS intervention.

• 4.4.1STEPS intervention delivery

  The STEPS intervention is delivered to individuals with ID residing in the group home and their residential staff as one group. In the intervention, there are six sessions delivered over 12 weeks and one booster session at 18 weeks, for a total of 7 sessions. In the sessions, we discuss the aspects of ADAPT. Interventionists have scripted modules for presentation of the materials, and the script is left in the home for staff. The scripted participant manual is aimed at second grade reading level and presents basic concepts with pictures to enhance comprehension of the concepts. The manual includes worksheets for practice between sessions.

• 4.4.2STEPS intervention content

  Participants are taught ADAPT model content in the following areas: in Session 1, participants are taught to work as a group, and the group process is facilitated by choosing a group name. In Session 2, participants are taught to stop and slow down, to recognize problems, and to be positive about being able to solve or manage the problems. In Session 3, participants learn how to define and break down problems. We cover questions such as whether they consider a problem big or small, whether they can change a problem or need to accept the problem and live with it, whether the problem involves one or many people. In Session 4, participants think about possible alternative ways to solve or deal with their problems and decide what they want to happen and whether that is reasonable. In Session 5, participants work on predicting possible outcomes and try out solutions. The material is cumulative, and each session builds upon the last. Sessions 6 and 7 are a review of all of the previous material. As in the pilot, the intervention sessions last approximately one hour and are generally conducted in the residences after dinner and before evening medications are handed out. Based on our previous work [21,22], at Sessions 2–6, we provide highlights of the previous sessions that are written by the project director and research assistant in consultation with the interventionists using a standardized format developed during our pilot. The highlights are brought to the following session to help with engagement and provide cues for retention of materials. The highlights are only for the residents and residential staff in the group home.

4.6. Attention-control: Food for life nutrition intervention

A nutrition intervention was chosen for the attention-control because nutrition is an issue among individuals with ID. The prevalence of overweight and obesity among adults with intellectual disabilities is higher than that of the general population and reportedly higher among individuals with ID living in community-based settings [59]. As with the STEPS intervention, the intervention is delivered to the individuals with ID residing in the group home and their residential staff as one group. As with STEPS, there are six sessions over 12 weeks, with a booster session at 18 weeks. Interventionist and participant manuals are standardized, and we cover aspects of My Plate [60] and recommendations from the American Heart Association on reduction of sugar [61]. At Sessions 2–6, we provide highlights of the previous sessions that are written by the project director and research assistant in consultation with the interventionists using a standardized format similar to STEPS highlights. For individuals with ID in both STEPS and attention-control Food for Life conditions, we are measuring weight and nutrition knowledge using the Nutrition and Activity Knowledge Scale, a measure specifically developed for individuals with ID [62,63]. For residential staff in both conditions, we are measuring nutrition knowledge and attitudes using two subscales of the USDA Diet and Nutrition Knowledge Survey [64,65].

4.7. Intervention staff

Intervention staff include data collectors, IFIRS problem-solving discussion videography team, interventionists, research assistants for the intervention, and assistants who rate the Problem-solving Task, the GEIS, and Breitenstein’s Fidelity Checklist from audiotapes.

• 4.7.1.Data collectors

  As in the pilot study [21,22], social workers or nurses with experience interview individuals with ID. This is to ensure high-quality data and to have staff able to address any mental health issues that arise during administration of the data collection tools, which include the Glasgow Depression Inventory. Data collectors are separate staff from staff conducting the intervention. The same data collector gathers measurement data for both the intervention and attention-control conditions. Every attempt is made to conceal which condition the participant is in from the data collectors [66].

• 4.7.2.IFIRS problem-solving discussion videography team (See Section 3.2.3). This team includes a research assistant who facilitates a problem-solving discussion and a videographer.
• 4.7.3.Interventionists

  Different from our pilot, for sustainability, and with consultation from our advisory group and others, we determined to use interventionists with the qualifications of Qualified Intellectual Disability Professionals, not psychiatric nurses. Qualified Intellectual Disability Professionals are required to have a bachelor’s degree in a human service field, with at least one year of experience working with individuals with ID; they typically provide service coordination and case management for residents with ID [53]. Interventionists are assigned to either the STEPS intervention or the attention-control Food for Life intervention, to minimize cross-over of the interventions.

• 4.7.4.Research assistants are graduate nursing students or assistants with experience in other research. They assist with the interventions by conducting such tasks as bringing materials, assisting with hand-out of highlights, and collecting satisfaction surveys. They are also assigned to either the STEPS intervention or the attention-control Food for Life intervention, to minimize cross-over of the interventions. Research assistants also assist with coding the Problem-Solving Task, the Group Environment for the Intervention Scale, and the Fidelity Checklist.

4.8. Fidelity plan

Treatment fidelity is assessed using the Behavior Change Consortium model [67], which assesses design, training, delivery, receipt, and enactment. As well, we modified Breitenstein’s Fidelity Checklist [68] to assess adherence to the STEPS and attention-control Food for Life interventions and competence in delivery.

• 4.8.1Design

  The intervention has been standardized with scheduled sessions and manuals outlining all session activities and length, facilitating maintenance of fidelity.

• 4.8.2Delivery

  Delivery is assessed. All 7 sessions for STEPS and the Food for Life attention-control are digitally audiotaped. We randomly select 25% of session audiotapes for research staff to score with Breitenstein’s Fidelity Checklist and observe in-person 10% of intervention sessions. This information is used to update training of research staff and prevent drift.

• 4.8.3Receipt

  Receipt is assessed in two ways. Attendance at the 6 sessions and 1 booster session is tracked. Attendance rates for individuals with ID and residential staff are calculated as the number of sessions attended divided by total number of sessions, based on attendance taken at each session. Reasons for not attending, if known, are collected throughout the intervention from individuals with ID and residential staff, and a log is kept. In addition, using a checklist we developed in our pilot study, we randomly select 25% of audiotaped sessions to determine the number of times each individual participated in discussion of a covered skill and whether the individual gave an example of how he/she used or would use the skill.

• 4.8.4Enactment is assessed by counting the number of completed and returned practice worksheets (total = 8 per individual with ID). We are writing an amendment to interview residential staff who participated in the intervention and administrative staff who assisted in setting up the intervention from our first two cohorts of homes about their involvement with the intervention, what they think they learned/gained from it, and how easy/hard it was for them to participate.

4.9. Training

The PI and the Project Coordinator (a social worker) train all interventionists, intervention research assistants, and data collectors. All are trained how to deal with adverse events, abuse, high depression scores, respond to suicidal intentions, and defuse and/or manage A/CBs displayed during group sessions or data collection. Standardized training manuals have been developed for (1) interventionists and intervention research assistants; (2) data collectors; and (3) problem-solving task raters. Interventionists and intervention research assistants in both conditions receive four hours of initial training and ongoing training in delivering either the STEPS or Food for Life intervention. Interventionists usually serve as research assistants for a few sessions before taking over the intervention. Members of the research team meet separately with STEPS interventionists and the Food for Life interventionists and research assistants to discuss any implementation issues. Data collectors receive four hours of initial training and ongoing training. The videography team is trained in the videotaping procedures for the Iowa Family Interactions Rating Scales. All training and supervision of data collectors and video data collectors is conducted separately from that of intervention staff. For staff coding the Problem-solving Task, the Group Environment for the Intervention Scale and Breitenstein’s Fidelity Checklist (all coded from audiotapes), training takes place with pre-scored audiotapes until at least 80% congruence with scores is achieved, and 15% of scores are checked for interrater reliability. Training of staff is ongoing to reinforce procedures. We keep and review notes of the training sessions and of research team meetings about training.

5. Data Analytic Plan

5.1. Data analysis

Data are entered into the REDCap database. Data management and analyses are conducted using SPSS for Windows (v. 24) and SAS (v. 9.4). Descriptive statistics for all variables are obtained, and t tests and chi-square analyses are performed on demographics, baseline life events, environment, and current health (depression symptoms and medication management) to verify that intervention and control groups are comparable. Missing data are imputed using SAS software using the multiple imputation strategy described by Rubin [69]. While we expect that all of the outcome measures will be close to normally distributed based on previous research, Tukey’s ladder of transformation [70] will be used to achieve normality as needed. If an outcome measure cannot be successfully transformed to achieve normality, it will be analyzed separately using the generalized multilevel analysis available in SAS.

• 5.1.1Specific aim 1 is to assess the efficacy of the STEPS intervention in group homes to improve social problem-solving skills and reduce A/CBs of individuals with ID compared to the attention-control Food for Life nutrition intervention. Because multiple dependent variables are being examined, a repeated-measures multivariate analysis of variance is being conducted to ensure control of experiment-wise α and to perform an initial assessment of intervention effects [71]. Intervention efficacy on individual measures of SPS skills and A/CBs of individuals with ID are calculated using a multilevel model with assessment. This assessment is nested within individuals nested within group homes. All analyses are performed on an intent-to-treat basis. Intervention condition is dummy-coded (treatment vs. control) at the level of the group home. To control for the effects of background characteristics on the intervention, measures associated with background variables are entered into the model as two propensity measures: one based on individual measures, and one based on group home measures. We use a two-level propensity analysis, a well-validated statistical approach useful in balancing groups on observed covariates, so that analyses of treatment effects are more accurate [72]. If a propensity measure has a nonsignificant effect, it is discarded; if significant, it is included in the final model. Categorical covariates (e.g., gender, ethnicity) are dummy-coded and entered into the model as a block. Propensity scores are estimated using individual participant (e.g., age, length of residence) and group home level measures (e.g. location [urban or suburban] and number of residents in the home) that could influence outcomes. We also check for differences in baseline SPS and A/CBs by agency and by interventionist; when found, we enter these into the propensity score analysis. This final set of variables (the direct predictors) constitute an optimal model for assessing each outcome measure.
• 5.1.2Specific aim 2

  Aim 2 is to assess the mediating effect of the support environment for SPS (residential staff SPS skills, group home level SPS skills, and group cohesiveness) on the improvement of SPS skills and reductions in the A/CBs of the individuals with ID. We expect the support environment for SPS (residential staff SPS, group SPS, and group cohesiveness) to mediate improvements that occur in the individuals with ID. To assess the mediating effects, we are modifying MacKinnon’s [73] regression models, which generate three regression equations and then examine the impact of a single mediator in terms of reduced variance explained in the outcome from the direct predictor variable when the mediator variable is included in the model. The same approach is used in this analysis, except that we examine multilevel regression instead of simple regression models [74,75]. With this modified approach, we estimate three sets of regression equations for each of the three potential mediators making up the support environment for SPS, crossed with each measure of SPS skills and A/CBs of individuals with ID. Intervention conditions are included as predictor variables in these mediation models. The first set of regression equations estimates the simple effect of the intervention on the SPS skills and A/CBs of the individuals with ID (Aim 1). The second set estimates the effect of the intervention on the mediator variables (residential SPS, group SPS, or group cohesiveness). The third set estimates the combined effects of the intervention and mediators on the group measures of SPS skills and A/CBs of the individuals with ID. We examine the impact of the change in the estimated coefficients associated with the introduction of each mediator. Significant mediators are left out of subsequent models because of concerns about collinearity. Retention of a significant mediator may make it difficult to detect mediation effects in subsequent analyses. Though several mediation models are being estimated in this aim, the analysis is based on the assumption that significance values are of limited use in analysis to find mediation effects [76]. Our analyses allows us to evaluate whether the support environment for SPS mediates the relation between the intervention and the outcomes and should give us information about which mediation variables are responsible for the effect [77,78].

• 5.1.3Specific aim 3

  Aim 3 is to evaluate the cost-effectiveness of STEPS relative to usual care for A/CBs incidents in group homes. This includes the cost from the societal perspective, taking into account costs borne by the program, group home, health care system, public services, participant, and family. The STEPS intervention is compared to usual care (control group costs of A/CBs excluding costs of the nutrition intervention). We conduct additional analyses separately from the perspective of the program, group home, health care system, public services, and participant. For cost measurement, the quantities of resources used and their associated prices are collected for the program (either prices paid or value of residential and administrative staff time), group home (value of residential and administrative staff time) and participant (value of participant’s time to participate in STEPS). For the effectiveness measurement, effectiveness is measured using the number of A/CB incidents. We also quantify the cost of each A/CB incident to calculate the net cost (or savings) of STEPS relative to standard of care. A/CB costs include for group home, participant and family members, public service (value of police officer time, ambulance), and health care system (cost per hospital emergency room visit, urgent care or other physician visit, hospitalization) costs. To calculate total program-related costs, the program, group home, and participant costs are summed to calculate total cost per participant. Similarly, total A/CB costs are summed across the group home, participant and family, public service and health care system. All costs are valued in 2016 dollars. Data for the cost-effectiveness analysis are drawn from study records and incident reports, which include such issues as how many phone calls were necessary, if staff meetings were necessary, if physicians were contacted. Cost-effectiveness is evaluated by combining the mean total program-related cost per participant with the number of A/CBs. We calculate the incremental cost-effectiveness ratio (ICER) for STEPS relative to usual care, such that ICER = (C1 − C0)⁄(E1 − E0), where C is program cost and E is effectiveness. Subscript 1 denotes STEPS and subscript 0 denotes usual care. 95% confidence intervals for the ICERs will be calculated to evaluate the uncertainty of these results [77–79]. We conduct one-way and multi-way sensitivity analyses for the key parameters to evaluate whether the ICERs are sensitive to plausible changes in their values. The sensitivity analysis is a check on the robustness and determines the key parameters affecting the ICERs. We also plot acceptability curves based on varying threshold (willingness to pay) values for a one-incident reduction in the number of A/CBs. Also, because A/CB incident costs are collected, we calculate the net cost (or net savings) associated with the STEPS intervention, such that the net cost = program-related cost – A/CB incident cost.

6. Limitations and Challenges

Potential limitations in the clinical trial include:

1. Although we are assessing the mediation effects of the support environment for social problem-solving in our analysis, we are using an attention-control condition rather than an assessment-only control, which may affect our findings.

2. The attention-control group may also be affected by being in a group intervention. To minimize this, we trained the interventionists for the Food for Life nutrition intervention separately from the STEPS interventionists. In their training, these interventionists focused exclusively on the nutrition content and not on group process.

3. The assessment process may affect the social problem-solving skills of residents and residential staff and the A/CBs exhibited by residents. To minimize this possibility, no feedback is given on responses to the data collection measures, and problem-solving is not practiced in the assessment process.

4. Incident reports are filled out by residential staff and reviewed by administrative staff. Bias in reporting by residential staff based on intervention condition is possible. However, the potential for bias is mitigated because community agencies have preexisting internal protocols for reporting incidents. During the initial discussions with agency administrators and residential staff participants, we explicitly instruct them to continue to use their existing protocols. In addition, we are collecting data on behaviors from non-residential (day or work program) staff where informants are naïve to study conditions.

5. We have chosen to randomize the homes prior to recruitment so that individuals with ID can be told about the intervention to which they are assigned. While randomization after recruitment and baseline data collection would be ideal, having to explain both conditions and the idea of random assignment to a condition would likely be confusing for adults with ID.

7. Discussion

The clinical trial of the STEPS intervention is an ongoing evidence-based, preventive intervention to decrease aggressive and challenging among people with ID. This intervention is an important step because previous studies using social problem-solving interventions with people with ID were conducted in clinical and forensic settings. In the STEPS project, adults with ID and staff in small community residences receive the intervention together in the home. By systematically involving residential staff in this intervention, we are able to test the mediating effects of the STEPS intervention on improvement in social problem-solving and decrease in A/CBs for both individuals with ID and the home as a whole. We are also able to look at the impact the STEPS intervention has on group cohesiveness. If the clinical trial data validate our hypotheses, the STEPS intervention will be an efficient, innovative, and cost-effective preventive intervention that will successfully reduce A/CBs for people with ID in small community group home settings. In addition, this intervention has the potential to reduce the public health impact that these behaviors have on individuals with ID and their caregivers.