Effects of a behaviour management technique for nursing staff on behavioural problems after acquired brain injury

ABSTRACT

The ABC method is a behaviour management technique for use by nurses. ABC refers to the identification of Antecedent events, target Behaviours, and Consequent events. In this longitudinal intervention study with double baseline measurements we evaluated the effectiveness of the ABC method in patients with behavioural problems after acquired brain injury. Fifty-six patients participated in this study. Outcome was measured in terms of overall neuropsychiatric problem behaviour, aggression, apathy and emotional burden experienced by nurses. A process evaluation was performed to investigate usability and acceptability of the method and identify factors that influenced effectiveness. Friedman’s ANOVA showed a small significant reduction over time in overall neuropsychiatric problem behaviour and aggression. The reduction was most prominent between the first baseline measurement and the post-implementation and follow-up measurements, not between the second baseline measurement and the post-implementation or follow-up measurements. This first group study on the effects of the ABC method could not prove this technique is effective for patients with behavioural problems after acquired brain injury. Nurses indicated that the ABC method was not fully implemented in their daily routines. This may have influenced results and makes it yet premature to draw firm conclusions on the effects of the ABC method.

KEYWORDS:

• Acquired brain injury
• behaviour management
• nursing staff

Introduction

Behavioural and emotional problems are common consequences of acquired brain injury (ABI), which put a high burden on the patient, their family and health care professionals (Alderman, 2007; Azouvi et al., 1999; Johnson & Balleny, 1996; Kant, Smith-Seemiller, & Zeiler, 1998).

Frequently identified problems after ABI are irritability, aggression, diminished anger control, depression and apathy. Prevalence rates for irritability and aggression after traumatic brain injury (TBI) for example range from 29% to 96% (see for example McNett, Sarver, & Wilczewski, 2012; Rao et al., 2009; Sabaz et al., 2014; Visscher, Van Meijel, Stolker, Wiersma, & Nijman, 2011). Scholten et al. (2016) in a systematic review found that prevalence rates for depressive disorders after TBI are up to 21% to 57% after follow-up periods of on average 1.5 to 3 years. Apathy has been reported for 46% to 71% of adults with TBI and for 15% to 71% of adults with stroke (see for example Caeiro, Ferro, & Cosat, 2013; Lane-Brown & Tate, 2009; Van Reekum, Stuss, & Ostrander, 2005). Reported prevalence rates of behavioural and emotional problems after ABI vary. The height of the reported rates depends in part on the severity of the injury of the patients who participate in the particular study and on the time of measurement. In addition the many different approaches that are used to identify behavioural and emotional problems also account for the wide variability in reported prevalence rates.

Behavioural and emotional problems may worsen over time (Alderman, 2003; Winkler, Unsworth, & Sloan, 2006). This may be a combined result of neuropathology and disappointments, frustrations and experiences of failure associated with poor performance in school or at work, social isolation and activity restrictions (McLean, Dikmen, & Temkin, 1993).

The persistence and magnitude of social and behaviour problems after ABI underline the need for effective interventions. A rich variety of behavioural management applications is available and has been applied in a variety of settings. Behavioural management applications used in brain injury rehabilitation derived largely from operant learning theory. An operant theory of learning is based on the concept that behaviour operates on the environment and is maintained by its consequences. The probability that any behaviour will occur again depends on whether it is rewarded/punished or not. An operant model provides a conceptual framework for understanding how reinforcement contributed to the development and maintenance of challenging behaviour, making it easier to develop effective interventions by creating a link between behaviour analysis and treatment (Wood & Alderman, 2011). The application of behaviour management for seriously challenging behaviour after brain injury has been described in the literature over several decades (see for example Wood & Eames, 1981).

In their systematic review on the effectiveness of behavioural interventions for individuals with behaviour problems after TBI, Ylvisaker et al. (2007) distinguished two main categories of intervention procedures: contingency management procedures, that is, the deliberate manipulation of consequences to encourage or discourage specific actions, and proactive antecedent-focused procedures, methods that manage immediate antecedents of behaviour. Examples of contingency management procedures are positive reinforcement, negative reinforcement, extinction and punishment (Alderman & Wood, 2013; Ylvisaker et al., 2007). Examples of pro-active antecedent-focused procedures are specifically planned tailor-made environmental structuring, provision of meaningful and well-understood daily routines and proactive development of positive communication alternatives to negative behaviour. Pro-active antecedent focused procedures often form part of a multicomponent intervention, which may also include contingency management procedures (Ylvisaker et al., 2007).

Recent reviews testify to the general effectiveness of behavioural interventions on patients with problem behaviour after ABI (Wood & Alderman, 2011; Ylvisaker et al., 2007). Most of the studies in these reviews reported positive results, despite the fact that the subjects of the study had often exhibited the disturbing behaviour for years. It has to be noted however that interpretation of these positive findings is clouded by the small number of RCTs and overreliance on evidence from single case studies, by the fact that failed interventions may not be published, and by a possible subject selection bias and hence a lack of knowledge regarding generalisation of results (Alderman & Wood, 2013; Wood & Alderman, 2011; Ylvisaker et al., 2007).

Members of nursing staff who are the main care providers for this patient group often feel that they lack the necessary skills to apply behavioural interventions in routine clinical practice, not having been sufficiently trained in the use of behaviour management techniques (Alderman, 2001; Mc Millan & Oddy, 2001; Wood & Alderman, 2011). Nevertheless, the role of the nursing staff is essential in reducing patients’ challenging behaviour, given that the vast majority of incidents of aggression, for example, are immediately preceded or followed by interactions between patients and nursing staff (Alderman, 2007; Pryor, 2004; Visscher et al., 2011). In fact, the nursing staff may unwittingly cause and/or maintain the challenging behaviour exhibited by patients (Wood & Alderman, 2011).

As early as 1994, Cohn, Smyer, and Horgas (1994) already recognised the important role of the nursing staff in reducing disruptive behaviour in geriatric patients. They developed the ABC method, a basic behaviour management technique designed especially for nursing staff. ABC focuses on the treatment of well-identified daily problem behaviour. The acronym ABC refers to the identification of Antecedent events or triggers, target Behaviours, and Consequent events. The method provides nurses with a framework in which they learn that their own behaviour and other factors (both internal factors and factors in the patient’s environment) can affect the emergence and persistence of challenging behaviour by patients. The principal idea is that situational and internal events trigger or influence the frequency and intensity of problem behaviours. The nurses learn to observe overt behaviour systematically, identify the triggers and use this insight to develop a clearly defined and feasible intervention plan to manipulate both consequences as well as immediate antecedents of behaviour to reduce undesirable behaviour and promote desirable behaviour. A major difference with other therapy interventions is that nursing staff is fully responsible for the application of the ABC intervention. This makes it different from other behavioural interventions, where a psychologist is often primarily responsible for designing and evaluating a behavioural therapy programme.

The ABC method was originally developed to reduce disruptive behaviour in geriatric patients. It seems likely that the method is also suitable for use in other groups of patients with cognitive impairments and problematic behaviour, as may be the case in patients with ABI. Although the ABC method has become popular in clinical practice, systematic evaluation studies on the effects are lacking. The primary objective of our study was therefore to evaluate the effects of this method on patients with problem behaviour after ABI. We hypothesised that the frequency and severity of overall neuropsychiatric problem behaviour and of aggression and apathy in particular would decrease after use of the ABC method by the nursing staff. In addition we hypothesised that use of the ABC method would lead to a reduction in the emotional burden experienced by nurses. To identify factors that can influence the effectiveness of the ABC method and to assess the usability and acceptability of the intervention, we performed a process evaluation.

Methods

Participants

Participants were recruited between August 2011 and May 2013. Participants were patients with ABI who were admitted to rehabilitation centre Adelante (Hoensbroek, the Netherlands) or to one of three Dutch permanent stay departments for patients with neuropsychiatric problems (SVRZ Ter Poorteweg Koudekerke, Department of Acquired Brain Injury Huize Padua GGZ Oost Brabant and Hambos clinic Kerkrade).

Inclusion criteria were: (1) diagnosis of non-progressive ABI (e.g., stroke or TBI) based on medical records; (2) ≥18 years of age; and (3) sufficient command of Dutch, based on clinical judgment. Exclusion criteria were: (1) minimally conscious state or post-traumatic amnesia at the time of recruitment and (2) no informed consent.

Design

We conducted a longitudinal intervention study with double baseline measurements. In our opinion, conducting a study with a (randomised) controlled design was not possible for several reasons. First, our intention was to train entire staff teams in using the ABC method and it was expected that once nurses were trained, they could not “switch off” the learned skills. As such, assigning half of the patients on a ward to an experimental ABC group with nurses using the ABC method, and the other half of the patients to care as usual with nurses not using the ABC method was not possible. Second, we felt that the patients from the four participating departments were not comparable to each other in terms of patient characteristics (e.g., type of injury, age) and living conditions (e.g., clinical rehabilitation centre versus permanent stay departments). And so randomising the four participating departments into two experimental groups and two control groups was not desirable. Instead, we chose to conduct a longitudinal intervention study with double baseline measurements. The rationale behind the double baseline measurements was including a “control” baseline measurement to check for spontaneous decreases in problem behaviour in the baseline data. We expected to find a decrease in problem behaviour over time, and that this decrease would be strongest after implementation of the ABC method on the wards, i.e., during the post-implementation and follow-up measurements. By monitoring a possible spontaneous decrease in problem behaviour during the baseline period, we would be able to check whether problem behaviour decreased spontaneously during the baseline period, whether problem behaviour decreased further after implementation of the ABC method on the wards, and whether decrease in problem behaviour was most significant between the second baseline measurement and the post-implementation or follow-up measurements rather than between the first baseline measurement and the second baseline measurement or between the first baseline measurement and the post-implementation and follow-up measurements. In that case we would conclude that the decrease in problem behaviour was due to implementation of the ABC method on the wards.

Procedure

The treating psychologist confirmed the eligibility criteria for all patients. Eligible patients were approached with written and oral information about the study. For those who could not read the information letter or understand the oral information (e.g., due to cognitive deficits), and could not decide upon participation, a legal representative was approached. The legal representative then decided on the patient’s participation in the study. After participants or their legal representative had provided written informed consent, the treating psychologist (or a trained intern supervised by the treating psychologist) retrieved demographic and clinical data from the medical records. Cognitive test were conducted at baseline by the treating psychologist, the principal investigator or a trained research assistant. Assessment of problem behaviour and burden experienced by nurses took place twice during the baseline phase (to control for spontaneous changes in problem behaviour in the baseline data), viz. at six weeks before introduction of the ABC method on the wards (b1) and immediately before its introduction (b2). Immediately after the baseline assessments (within one week after b2), ABC training was provided (three sessions spread over four weeks), and the method was introduced on all wards. From the first training session on, nurses were urged to continue using the method beyond the follow-up measurements. Post-implementation assessment of problem behaviour and the burden experienced by nurses took place within one to two weeks after the last training session (p1, i.e., six to seven weeks after b2) as well as nine weeks after p1 (follow-up measurement, fu). A process evaluation was conducted immediately after the follow-up measurement.

This study was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki). The Medical Ethics Committee of the Maastricht University Medical Centre and all participating departments approved the study protocol. All patients (or their legal representative) gave informed consent.

Treatment method

All nurses who worked with the patients were asked to attend a three-day training course (three sessions spread over four weeks) with a total duration of 15 hours. Nurses were trained in working according to the ABC method developed by Cohn et al. (1994) and translated into Dutch by Hamer and Voesten (2001). The original method focused on dealing with behavioural problems in geriatric populations. For the purpose of this study, we shortened the course from six to three days and from six to five modules (we left out the module in which participants apply the ABC method on themselves and their own situation/environment). We also slightly adjusted the method for use in an ABI population, which meant that we used practice tasks that focused on younger patients.

The ABC method rests on two main pillars. The first focuses on communication. Since most aggression incidents are preceded by interactions between patient and nurse, it is essential that communication between the two is optimal. The method provides nurses with information, advice and guidance for both verbal and non-verbal communication.

The second pillar of the ABC method focuses on a basic behaviour management technique. The A, B and C are the three main steps of both the observation and treatment/change phases. In each phase, all nurses jointly formulate questions that describe the antecedents (A), the challenging behaviour (B), and the consequences (C) of the behaviour.

Key items are observation and analysis of the challenging behaviour. Nurses observe the challenging behaviour (by watching the patient, listening to the patient, etc.). To prevent nurses from jumping to conclusions, they are asked to list up all possible internal and external antecedents of the behaviour. There can be many causes of or reasons for the observed problem behaviour. Restless behaviour for example (such as roaming about or repeatedly asking the same question) may have medical causes (e.g., cognitive deficits such as memory or orientation problems, a urinary infection, or known lower back problems), but may also have emotional/psychological causes (such as need for attention, feelings of boredom); and there may be environmental factors (too much light or noise, too few directing stimuli). Apathetic behaviour may have medical causes (a brain injury-related pseudo-depression for example) or psychological/emotional causes (e.g., feelings of unsafety, few positive experiences, too many negative experiences) to name just a few. And last, agitated behaviour may, among other causes, have environmental causes (a new client or employee who talks a lot, sits too close to the patient, etc.)

The three steps of the observation phase automatically introduce “three steps for changing behaviour”: what kind of behaviour would we like to achieve, how can the antecedents be changed and how can the consequences be changed? These questions form the basis for a problem-oriented treatment plan, in which well-defined, concrete and observable desired “new” behaviour is formulated as a substitute for the undesirable challenging behaviour (e.g., “does not curse at nursing staff during ADL care in the morning”). Antecedents that lead to problem behaviour and that can be taken away, and alternative antecedents or triggers that can encourage the desired behaviour are clearly described. Finally, the expected new and positive consequences are described. The list of antecedents is reviewed, and the antecedent that is judged by the nursing team as the most likely cause of the problem behaviour (and that can be influenced) is targeted first. If the intervention is not successful, a further review of antecedents and consequences is undertaken and a new plan for change, targeted at the antecedent that is the next most likely cause of the problem behaviour, is formulated.

Chosen interventions may consist of reinforcing desirable behaviour by rewarding it (e.g., a patient who values table manners is allowed to eat at a separate table, or a patient who is apathetic/depressed because he is feeling more and more disabled is given subtasks that are known to be manageable by the patient, to increase the frequency of positive experiences). Interventions also may entail reinforcing desirable behaviour by taking away a negative experience for the patient (e.g., not seat a patient who is agitated easily next to a very talkative patient), or discouraging undesirable behaviour by ignoring it (e.g., the nurse leaves the room for a minute). Choice of type of intervention depends on a patient’s capacities. For patients who have deficits in reward sensitivity or who show severe cognitive deficits, interventions that rely on ignoring undesirable behaviour may be more successful than interventions that rely on positive reinforcement. Table 1 illustrates the three steps of observing behaviour and the three steps for changing behaviour.

Table 1. Steps in observing behaviour and steps for changing behaviour.

Three steps in observing behaviour:

Antecedent event (A): describe possible antecedent stimuli (triggers or situations). What is going on in the patient him/herself? What is happening in the patient’s immediate environment? Has anything changed recently?

Target Behaviour (B): describe the behaviour clearly. What is the patient doing? Where/when does it occur? How long has this been going on? How often does it occur and how serious is it? For whom is it a problem?

Consequent events (C): describe the consequences. How does the behaviour continue? What is the response of the nurses and others to the patient’s challenging behaviour? What are the consequences for the patient?

Three steps for changing behaviour:

New Antecedent events (A): Which old antecedents can be altered? Which new antecedents could encourage the desired behaviour?

Desired Behaviour (B): What do we want the patient to do? Is it possible to achieve this all at once? Is the desired behaviour clearly described? Is it attainable?

New Consequent events (C): Can we avoid old consequences? Which new consequences stimulate the desired behaviour?

Measurements

Demographic and clinical patient characteristics: The following information was extracted from the patients’ files: gender, age, type and date of brain injury, and psychiatric history.

To be able to give a broad description of patients’ overall cognitive and frontal functioning, cognitive tests were conducted at the first baseline measurement. Patients’ overall cognitive functioning was evaluated with the Mini Mental State Examination (MMSE) (Folstein, Folstein, & McHugh, 1975). Frontal functioning was evaluated with the Frontal Assessment Battery (FAB) (Mungas, 1991) and the Key search test of the Behavioural Assessment of the Dysexecutive Syndrome (BADS) (Wilson, Alderman, Burgess, Emslie, & Evans, 1996).

Primary outcome measures

Overall neuropsychiatric problem behaviour: The Neuropsychiatric Inventory (NPI) was used to investigate patients’ overall behavioural and emotional problems (Cummings et al., 1994). Frequency and severity scores were recorded. The overall NPI score ranges from 0 to 144 (higher scores meaning more frequent and more severe problem behaviour).

Apathy: The Apathy Evaluation Scale (Lampe, Kahn, & Heeren, 2001; Marin, Biedrzycki, & Firinciogullari, 1991) was used to investigate apathy. This scale consists of 18 items (scored on a 4-point scale), with total score ranging from 18 to 72 (higher scores meaning more frequent and more severe apathy).

In addition the apathy subscale of the Neuropsychiatric Inventory (NPI) was used to investigate the frequency and severity of patients’ apathetic behaviour (Cummings et al., 1994). NPI-apathy scores range from 0 to 12 (higher scores meaning more frequent and more severe problem behaviour).

Aggression: For aggression several primary outcome measures were used, that all measure slightly different aspects of aggressive behaviour. The Staff Observation Aggression Scale-Revised (SOAS-R) was used to document the occurrence, nature and severity of separate aggression incidents (Nijman et al., 1999). The instrument is used to record triggers that led to the aggressive behaviour, means that were used by the patient during the incident, the target of the aggressive behaviour, consequences for victims, and measures taken to control or stop the aggression. The overall severity score per incident ranges from 0 (least severe) to 22 (most severe aggression incident). In our study, the overall severity scores of all incidents that occurred on seven consecutive days were added up and used as the total SOAS-R score.

The Agitated Behaviour Scale (ABS) was used to measure the severity of agitation (Corrigan, 1989). The minimum total score is 14, the maximum score 64 (high scores representing severe agitation). In our study, a nurse completed the scale every day for seven consecutive days, at the end of an eight-hour shift. The mean score over seven days was used as the overall ABS score.

The Social Dysfunction and Aggression Scale (SDAS-11) was used as an observation scale to document a nurse’s overall impression of a patient’s aggressive behaviour during a certain time period (Wistedt et al., 1990). The total score ranges from 0 to 44 (higher scores representing more severe aggressive behaviour).

In addition the aggression subscale of the Neuropsychiatric Inventory (NPI) was used to investigate the frequency and severity of patients’ aggressive behaviour (Cummings et al., 1994). The NPI-aggression score ranges from 0 to 12 (higher scores meaning more frequent and more severe problem behaviour).

All primary outcome measures were completed by the nurses.

Secondary outcome measures

Burden experienced by nurses: The NPI was used to investigate nurses’ emotional burden due to patients’ neuropsychiatric problem behaviour (Cummings et al., 1994). For each neuropsychiatric symptom, nurses scored the severity of the emotional burden they experienced, from 0 (no burden) to 5 (severe or extreme burden). Total burden scores range from 0 to 60 (higher scores representing more severe burden experienced by nurses).

Process evaluation

All nurses who had participated in the ABC training course individually completed an evaluation form (usability and acceptability questionnaire) after the follow-up measurement. Nurses indicated whether they thought the ABC method was instructive and educational, their opinion on effectiveness of the method, whether the method was being used on the ward after introduction, and whether the team cooperated in using the method. Scores on these questions ranged from 1 (totally disagree) to 5 (totally agree). In addition, there were open questions asking about bottlenecks and prerequisites for implementation.

After completion of the individual usability and acceptability questionnaires, the nurses were invited for a semi-structured group interview to further reflect upon bottlenecks and prerequisites for implementation of the ABC method on their ward (e.g., time investment and team cooperation).

Statistical analyses

Demographic and injury-related characteristics: Descriptive statistics were used to describe demographic and injury-related parameters (gender, age, educational level, type of injury, time since injury, psychiatric history, overall cognitive and frontal functioning).

Effects on neuropsychiatric problem behaviour and burden experienced by nurses: Because the assumption of normal distributions was violated and transformations proved no solution because of the (rather strongly) skewed data, Friedman’s ANOVA analyses with post-hoc comparisons were used to study changes in the frequency and severity of overall neuropsychiatric problem behaviour and of aggression and apathy in particular, and changes in the burden experienced by nurses. Effect sizes were calculated using r values. The r value was considered small when between 0.1 and 0.3, moderate when between 0.3 and 0.5 and large when greater than 0.5 (Rosenthal & Rosnow, 1984). Results of the Friedman’s ANOVA analyses were considered significant if p < .05; for the post-hoc comparisons a Bonferroni correction was applied, and so post-hoc effects are reported at a (0.05/5) 0.01 level of significance. All statistical analyses were conducted using SPSS 20.0 for Mac OS X.

Process evaluation

The quantitative data from the usability and acceptability questionnaires were analysed with descriptive statistics using SPSS 20.0 for Mac OS X. Qualitative data resulting from the open questions on the usability and acceptability questionnaires and from the interviews with the nurses were clustered based on the contents of the answers.

Results

Demographic and injury-related characteristics

The study sample included 102 patients. Forty-two patients were discharged home before the double baseline measurements were completed. Three patients were transferred to other centres or departments before the post-implementation measurements. One patient had died before the follow-up measurements. At follow-up, data were therefore available for 56 patients. Table 2 shows the patients’ demographic and injury-related characteristics.

Table 2. Demographic and injury-related characteristics (n = 56).

Gender, n male	44
Mean age in years (SD; range)	61.0 (11.9; 34–84)
Type of brain injury, n
Contusion	4
Stroke	4
Encephalopathy	2
Korsakoff	19
Epilepsy	2
Other	2
Mixed*	12
Unknown	11
Mean time since injury in years (SD; range)	14.0 (11.2; 1.0–51.6)
Co-morbid psychiatric diagnosis, n	26
Mean MMSE score^a (SD; range)	21.3 (5.8; 3–30)
Mean FAB score^a (SD; range)	10.6 (4.1; 3–18)
Mean Key search profile score^a (SD; range)	1.6 (1.4; 0–4)

*n = 1 hydrocephalus and status after neurosurgery; n = 1 meningitis and stroke; n = 1 Korsakoff and encephalopathy n.s.; n = 2 Korsakoff and stroke; n = 1 Korsakoff and epilepsy; n = 1 Korsakoff and trauma; n = 1 Korsakoff and hydrocephalus; n = 2 Korsakoff and epilepsy and stroke; n = 1 alcohol abuse and contusion and epilepsy and encephalopathy n.s.; n = 1 alcohol abuse and contusion and stroke and meningitis.

^aMMSE = Mini Mental State Examination; FAB = Frontal Assessment Battery; Key search = Key search test of the Behavioural Assessment of the Dysexecutive Syndrome.

Most patients were diagnosed with Korsakoff’s syndrome (34%). Twelve patients (21%) had mixed diagnoses, most of whom (83%) were diagnosed with Korsakoff and one or more other illnesses such as hydrocephalus or stroke. For 11 patients (20%) the exact cause of the ABI was not specified. Twenty-six patients (46%) had comorbid psychiatric disorders (e.g., depression, psychosis). Fifty-four percent of the patients had disorders of overall cognitive functioning (MMSE-score < 24), and 45% had frontal executive disorders (FAB-score < 12; key search profile score < 2).

Neuropsychiatric problem behaviour

The means and standard deviations, medians and ranges of the outcome measures at baseline (b1 and b2) and at post-implementation (p1) and follow-up (fu) assessments are displayed in Table 3.

Table 3. Friedman’s ANOVA—neuropsychiatric behaviour outcome measures.

Measure	Domain	N	b1^b Mean (SD); Median (Range)	b2^b Mean (SD); Median (Range)	p1^b Mean (SD); Median (Range)	FU^b Mean (SD); Median (Range)	Χ^2-value	r value
NPI over all^a	Overall neuropsychiatric problem behaviour	50	23.28 (20.44); 18.00 (90.00)	21.50 (17.35); 19.50 (62.00)	19.24 (17.12); 15.00 (57.00)	17.96 (15.89); 17.50 (57.00)	11.65**	.17
NPI aggression^a	Aggression	55	3.04 (3.73); 2.00 (12.00)	2.49 (3.42); 0.00 (12.00)	1.91 (2.83); 0.00 (12.00)	1.56 (2.43); 0.00 (12.00)	14.29**	.18
SOAS-R^a	Aggression	52	4.40 (9.08); 0.00 (51.00)	2.06 (6.25); 0.00 (28.00)	1.42 (3.92); 0.00 (21.00)	2.52 (6.70); 0.00 (30.00)	21.13**	.23
ABS^a	Aggression	53	20.22 (3.69); 20.10 (15.10)	19.29 (4.57); 18.00 (18.40)	18.42 (3.29); 18.10 (14.70)	18.71 (4.27); 17.40 (18.60)	17.20**	.21
SDAS-11^a	Aggression	46	7.41 (6.71); 5.50 (21.00)	6.26 (6.63); 3.00 (20.00)	5.48 (6.14); 3.00 (21.00)	5.41 (6.18); 2.50 (22.00)	10.12*	.15
NPI apathy^a	Apathy	51	2.04 (3.50); 0.00 (12.00)	2.65 (3.43); 0.00 (12.00)	2.69 (3.63); 1.00 (12.00)	2.14 (3.31); 0.00 (12.00)	3.25	.02
AES^a	Apathy	53	36.96 (11.96); 35.00 (50.00)	36.64 (12.83); 36.00 (51.00)	36.51 (11.69); 34.00 (46.00)	35.77 (11.21); 35.00 (45.00)	1.54	.03
NPI burden^a	Burden experienced by nurses	55	6.80 (6.64); 6.00 (24.00)	7.21 (7.37); 5.00 (29.00)	6.36 (6.97); 3.00 (25.00)	6.44 (6.57); 5.00 (26.00)	1.24	.04

^aNPI: Neuropsychiatric Inventory; SOAS-R: Staff Observation Aggression Scale-Revised; ABS: Agitated Behaviour Scale; SDAS-11: Social Dysfunction and Aggression Scale; AES: Apathy Evaluation Scale.

^bb1 = first baseline measurement; b2 = second baseline measurement; p1 = post-implementation measurement; fu = follow-up measurement.

*p < .05; **p < .01.

Friedman’s ANOVA analyses showed small but significant time effects for overall neuropsychiatric problem behaviour and for each aggression measure: overall neuropsychiatric problem behaviour and aggressive behaviour diminished over time (NPI overall: Χ²(3, N = 50) = 11.65, p = .009, r = 0.17; NPI aggression: Χ²(3, N = 55) = 14.29, p = .003, r = 0.18; SOAS-R: Χ²(3, N = 52) = 21.13, p = .000, r = 0.23; ABS: Χ²(3, N = 53) = 17.20, p = .001, r = 0.21; SDAS-11: Χ²(3, N = 46) = 10.12, p = .018, r = 0.15). Apathy did not diminish significantly over time (NPI apathy: Χ²(3, N = 51) = 3.25, p = .36, r = 0.02; AES: Χ²(3, N = 53) = 1.54, p = .67, r = 0.03) (Table 3).

Post-hoc analyses showed that the reduction in overall neuropsychiatric problem behaviour and in aggressive behaviour was significant between the first baseline measurement and the post-implementation or follow-up measurements (NPI overall, ABS and SDAS-11), and between the first and second baseline measurement (ABS). Between the second baseline measurement and post-implementation or follow-up measurements no significant reduction was found (Table 4).

Table 4. Pairwise post-hoc comparisons for outcome measurements.

Measure	b1-b2^b z-value (r-value)	b1-p1^b z-value (r-value)	b1-fu^b z-value (r-value)	b2-p1^b z-value (r-value)	b2-fu^b z-value (r-value)
NPI overall^a	1.278 (.13)	2.789* (.28)	2.905* (.29)	1.510 (.15)	1.627 (.16)
NPI aggression^a	1.366 (.13)	2.585 (.25)	2.253 (.21)	1.219 (.12)	.886 (.08)
SOAS-R^a	2.241 (.22)	2.545 (.25)	1.747 (.17)	.304 (.03)	−.494 (.05)
ABS^a	2.282* (.22)	3.273* (.32)	3.687* (.36)	.451 (.04)	.865 (.08)
SDAS-11^a	1.638 (.17)	2.796* (.30)	2.277; (.24)	1.158 (.12)	.639 (.07)

^aNPI: Neuropsychiatric Inventory; SOAS-R: Staff Observation Aggression Scale-Revised; ABS: Agitated Behaviour Scale; SDAS-11: Social Dysfunction and Aggression Scale.

^bb1 = first baseline measurement; b2 = second baseline measurement; p1 = post-implementation measurement; fu = follow-up measurement.

*p ≤ .01.

To further control or analyse for variables that may have influenced results, additional post-hoc stratified analyses were done. First we checked whether there were differences in changes over time between the three different sites (Friedman’s ANOVA analyses per site) to control for any influences due to differences in staff (years of experience, additional training), differences in services, patient differences or differences in success of implementation across sites. At each site, behaviour decreased over time. For two sites (Huize Padua (n = 11) and Hambos (n = 7)), changes were not significant. For SVRZ, the post hoc analyses showed that the decrease was due to a decrease between the first baseline measurement and the post-implementation or follow-up measurements, not between the second baseline measurements and post-implementation or follow-up measurements. Second, we checked whether there were differences in changes over time between clinical groups. We compared the group of patients with Korsakoff’s syndrome (including patients with Korsakoff’s syndrome and one or more other illnesses, n = 27) to the group of patients with other diagnoses (n = 27) (Friedman’s ANOVA analyses per group). For some outcome measures both groups showed decreases in scores over time, for other measures the Korsakoff group showed significant decreases over time whereas the non-Korsafkoff group did not, and for still other measures the non-Korsakoff groups showed decrease over time whereas the Korsakoff group did not. When decrease over time was significant, the post-hoc analyses showed that the decrease was between the first baseline measurement and the post-implementation or follow-up measurements, not between the second baseline measurements and post-implementation or follow-up measurements. Third, we checked whether there were differences in changes over time between those patients with higher cognitive (MMSE > 23, n = 19) versus lower cognitive functioning (MMSE < 24, n = 29) and between those patients with higher executive (FAB > 10, n = 23; Key search test profile score 3–4, n = 16) versus lower executive functioning (FAB < 11 = 23 ; Key search test profile score 1–2, n = 30) (Friedman’s ANOVA analyses per group). For some outcome measures both groups showed decreases in scores over time, for other measures the “lower cognitive/executive group” showed significant decreases over time whereas the higher group did not, and for still other measures the higher group showed decrease over time whereas the lower group did not. When decrease over time was significant, post-hoc analyses showed that the decrease was between the first baseline measurement and the post-implementation or follow-up measurements, not between the second baseline measurements and post-implementation or follow-up measurements. Last, we checked whether there were differences in changes over time between those patients with additional psychiatric diagnoses (according to DSM-IV diagnosis or based on description of psychiatric history; n = 26) versus those without additional psychiatric diagnoses (n = 29). For some outcome measures both groups showed decreases in scores over time, for other measures the group with additional psychiatric diagnoses showed significant decreases over time whereas the other group did not, and for still other measures the group without additional psychiatric diagnoses showed decreases over time, whereas the other group did not. When decrease over time was significant, post-hoc analyses showed that the decrease was between the first baseline measurement and the post-implementation or follow-up measurements, not between the second baseline measurements and post-implementation or follow-up measurements.

Burden experienced by nurses

The means and standard deviations, medians and ranges of the NPI total burden scores at baseline (b1 and b2) and at post-implementation (p1) and follow-up (fu) assessments are displayed in Table 4. Friedman’s ANOVA analyses showed no significant reduction over time in the burden experienced by nurses (NPI burden: Χ²(3, N = 55) = 1.24, p = .74, r = 0.04) (Table 3).

Process evaluation

In total, 43 nurses completed the process evaluation form. Table 5 shows nurses’ mean scores on the usability and acceptability questionnaire. Nurses reported that learning to work with the ABC method was very instructive and educational; 98% of the nurses gave scores of 4 (agree) or 5 (totally agree). More than half of the nurses (65%) reported that teams were not yet using the ABC method as part of their routine clinical practice at the time of the follow-up measurement. As a consequence, half of the nurses (51%) could not yet reflect on their perception of the effectiveness of the method. Further, 77% of the nurses reported that the entire staff team was rarely involved in using the ABC method.

Table 5. Nurses’ mean scores (and SDs) on usability and acceptability of the ABC method (n = 43).

Instructiveness of ABC*	Effectiveness of ABC*	Use of ABC on the ward*	Team cooperation when using ABC*
4.42 (.63)	3.49 (.63)	3.07 (1.06)	3.00 (.87)

*Scores could range from 1 (totally disagree) to 5 (totally agree).

Lastly, nurses were asked to report bottlenecks and prerequisites for implementation of the ABC method on their ward. Several reasons were given as to why the ABC method was not part of routine clinical practice at the time of the post-implementation measurements. First, nurses still felt insecure about their expertise in using the ABC method; they felt that they still lacked the necessary skills and that they needed more practice to become confident about using the method independently. Second, nurses felt that they had no time during their shifts to sit down together and identify the steps for observing behaviour and the steps/agreements for changing behaviour. The time pressure they experienced meant that they did not give priority to working with the ABC method. And third, nurses felt that not every colleague was as yet convinced of the effectiveness of the ABC method and therefore felt no urge to become acquainted with it and make it part of their routine clinical practice. The most important prerequisite for implementation of the ABC method on the ward was having an “ABC ambassador”. Staff teams need to appoint a colleague who puts ABC on the agenda of team meetings, who includes every staff member in identifying the steps used in observing behaviour and agreements in changing behaviour, and who calls staff members who do not use the ABC method to account.

Discussion

The ABC method aims to identify concrete, well-defined and observable problem behaviour and triggers that can be changed easily by the nursing staff. The main aim of our study was to investigate whether the ABC method could effectively reduce problem behaviour after ABI.

The results showed a significant reduction over time in overall neuropsychiatric problem behaviour and aggression, but not in apathy. The reduction was, however, most prominent between the first baseline measurement and the post-implementation or follow-up measurements; reduction between the second “control” baseline measurement and post-implementation or follow-up measurements was not significant. This implies that the reduction may not be the result of treatment. There may have been an anticipatory effect of staff paying more attention to problem behaviour during the baseline measurements, which may have caused them to (intentionally or unintentionally) change their own behaviour, thereby already reducing patients’ disruptive behaviour during the baseline period. This may be interpreted as an example of the classical Hawthorne effect, which refers to a type of reactivity in which individuals improve an aspect of their behaviour in response to their awareness of being observed (Fox, Brennan, & Chasen, 2008; Mc Carney et al., 2007). Since our study sample had on average acquired their ABI many years ago, the reduction is unlikely to be the result of spontaneous recovery.

Our finding that the ABC method was not effective in this sample of patients is not in line with results from other studies on the effects of behavioural management techniques in patients with problem behaviour after ABI. In their review on effects of contingency management procedures and proactive antecedent-focused procedures, Ylvisaker et al. (2007) found that all of the studies in the review demonstrated improvement in at least one measured outcome. Our study differed in several ways from the studies in the review. First, it is possible that fewer of the patients in the review suffered from secondary psychiatric illnesses than in our sample. The authors of the studies included in the review frequently provided insufficient information on dual diagnoses such as substance abuse and psychiatric disorders. In many cases, co-existing impairments were not fully described. Differences in the presence of secondary psychiatric illness may explain the differences between the findings from the review and the findings from our study: it is possible that our patient group was more resistant to change due to psychiatric comorbidity. Second, treatment duration and frequency may have differed between our study and the studies described in the review. As regards frequency of intervention, it is difficult to compare our study with those in the review. Our study used an approach that is partly environmental in nature, with all relevant staff trained to implement the intervention throughout the day. Frequency and intensity of treatment cannot be observed. As regards treatment duration, the majority of interventions used in the studies in the review generally ranged from 1 to 6 months. In our study, follow-up measurements took place nine weeks after the last training session. Treatment duration in our study may thus have been shorter than that in most studies in the review. Third, behaviour management techniques like those described in the review and that used in our study aim at very specific problem behaviour. Standardised behaviour scales (such as those used in our study) may be less sensitive in picking up (subtle) changes in specific behaviour. Most of the studies in the review were single-subject experiments that analysed decreases or increases in behaviours that were specifically targeted by the intervention. Only a few studies included standardised behaviour scales. These differences between the studies included in the review and our study may explain the differences between the findings.

Our study included a process evaluation to examine the usability and acceptability of the ABC method and identify factors influencing its effectiveness. Nurses rated the method as instructive and educational, but also reported that it was not yet part of routine clinical practice at the time of the post-implementation measurements: the method was not used in every situation in which problem behaviour occurred, and rarely were all colleagues involved in using the ABC method. A major difference with other behaviour modification therapies is that members of nursing staff are put in charge of applying the ABC intervention. It is likely that nurses need more time to get used to taking responsibility. Also, nurses still felt insecure about their expertise in using the ABC method; they felt that they still lacked the necessary skills and that they needed more practice to learn these new skills and become confident about using the method independently. We shortened the original ABC training course from six to three days, with a total duration of 15 hours. This may have been too short a period for the nurses to feel secure about taking responsibility and about their skills. This may explain their hesitation to take action and explicitly recording the three steps of observation and the treatment plan. However, it is likely that the introduction of the method did change the way that they look at disruptive behaviour, and caused them to change their own behaviour in response (i.e., altering old antecedents without explicitly making an observation and change plan). The fact that not every colleague was involved in using the method may also have influenced the results. One of the greatest obstacles to the use of behaviour management techniques is the level of consistency required to achieve success. As described in the introduction section, these techniques most heavily rely on operant learning, which is based on the concept that behaviour is maintained by its consequences. The probability that any specific behaviour will occur again depends on whether it is rewarded or not (Wood & Alderman, 2011). The use of the ABC method leads to a clear set of procedures that need to be followed to ensure a consistent approach by staff. Even if a single staff member does not comply with the agreements made in the treatment plan, this can prolong a patient’s challenging behaviour. In our study we have not managed to educate each and every nursing staff member of every participating ward about the rationale of the ABC method and train them in working according to the ABC method (due to practical issues such as holidays of staff members). We did not evaluate in a quantitative manner how many nurses were trained and how many were not (and whether trained nurses missed training sessions), so we cannot give exact numbers. However we acknowledge that even though the staff members who did receive the training were urged to educate their colleagues, the fact that not every nursing staff member attended the (entire) training course may have influenced the results. Some more detailed information on this would have been useful to understand the potential impact on effects and success of implementation. Furthermore we acknowledge that a consistent approach by all the professionals, care staff and support staff that come into contact with patients is preferred (Wood, 2001). However, other medics and paramedics (psychiatrists, general practitioners, physiotherapists, etc.), i.e., all employees with care tasks but who come into contact with the patients at an irregular or infrequent basis, were not specifically trained in working with the ABC method. Housekeeping and kitchen personnel also were not trained in working with the method. Although these staff members have no specific care tasks, they often come into contact and communicate with the patients. Even without knowing, they may say or do something that does not comply with the agreements made in the treatment plan. This may have influenced results.

A second aim of our study was to investigate whether the use of the ABC method would lead to less burden experienced by nurses. No reduction in this burden was shown. One possible explanation for the lack of significant findings on burden may be that the mean baseline scores as well as those found post-implementation were in the average/non-distressed range, suggesting that there may have been a floor effect. Wade, Michaud, and Brown (2006) in their study on effects of a family problem-solving intervention for children with TBI found similar results for parental distress. In a more distressed sample, decreases in aggression might have been associated with greater improvements in nurses’ burden.

Limitations of the study

This study has several limitations. First, we acknowledge that conducting a single case experimental design study or a randomised controlled trial would be preferable when evaluating interventions like that used in our study. An important disadvantage of single case studies however, is that it is difficult to generalise results and to decide whether the studied method is usable/effective for other patients. This is one reason why we chose to perform a group study.

Subsequently, we felt that conducting a RCT with randomising the four participating departments into two departments with nurses using the experimental ABC method and two control departments with nurses not using the ABC method was not desirable. The patient groups differed in terms of department characteristics (some patients were referred to clinical rehabilitation whereas others lived in permanent stay departments) and patient characteristics (in one department most patients were young patients suffering from traumatic brain injury, in another department most patients were older stroke survivors, and in yet another department most patients were diagnosed with Korsakoff’s syndrome). In addition we felt that assigning half of the patients of one department to an experimental ABC group with nurses using the ABC method, and the other half of the patients of the department to care as usual with nurses not using the ABC method was not possible. Entire staff teams were trained in using the ABC method. And since they learned new skills and to change the way they respond to patients’ problem behaviours, it was expected that once nurses were trained, they could not “switch off” the learned skills.

As such, we chose to perform a longitudinal intervention study including a second “control” baseline measurement to check for spontaneous decrease in problem behaviour in the baseline data. We had expected to find a decrease in problem behaviour over time, and that this decrease would be strongest after implementation of the ABC method on the wards, i.e., during the post-implementation and follow-up measurements. Since the problem behaviour decreased strongest during the baseline period, and less so during the post-implementation and follow-up period, we cannot conclude that the ABC technique is effective for patients with behavioural problems after acquired brain injury. Additional post-hoc stratified analyses showed that results were comparable across settings (regardless of differences across sites in staff (years of experience, additional training)/ in services/ in patient groups/ in success of implementation), across diagnostic groups, across groups with different levels of cognitive/executive functioning, across groups with or without additional psychiatric diagnoses, and across groups with different levels of problem behaviour at baseline. The reduction over time in problem behaviour was most prominent between the first baseline measurement and the post-implementation and follow-up measurements; not between the second baseline measurement and the post-implementation or follow-up measurements.

A second limitation of this study is that 45% of patients were lost to follow-up for various reasons, leaving a small study sample. Third, the follow-up period may have been too short to yield significant treatment results. As referred to above, the nurses were in charge of working according to the ABC method, and it is very likely that they needed more time to get used to taking responsibility and become confident in independently using the method without supervision from a psychologist or psychiatrist. The ABC method is about learning new skills, not just gaining more knowledge, and learning new skills takes time. It may also take time to find the right treatment approach (i.e., the antecedents that encourage the desired behaviour). Several reviews of antecedents and consequences may be required before a successful intervention plan is formulated. A follow-up period of nine weeks may be too short to yield significant treatment results. We expect that a longer follow-up period with post-implementation assessments at 6 months or a year after implementation on the ward may show larger treatment effects. Fourth, it is possible that the problem behaviour was assessed by different nurses at the different measurement time points. This may have biased the results, as one nurse may experience the disturbing behaviour as more severe than another nurse.

Conclusions and recommendations for implementation and future studies

This first study on the effects of the ABC method for use by nurses could not prove that this method is effective for patients with behavioural problems after acquired brain injury. Interviews with the nurses who were trained to use the method showed that it was not fully implemented in their daily routines. This may have influenced results and makes it yet premature to draw firm conclusions on the effects of the ABC method. Further research into the effectiveness of the ABC method is warranted. For future studies we recommend using longer follow-up periods and using outcome measures aimed at the specific problem behaviours targeted in the intervention. Above all, as regards clinical practice, we recommend putting more effort into the implementation of the method on the ward. First, enough practice is needed for nurses to get acquainted with the new skills, to take away their insecurity and improve their confidence in using the method. Second, before introducing the method on the ward, (financial) management should be involved and agreements should be made that nurses are given time to sit down together in peer-to-peer coaching groups to identify the steps for observing behaviour and the steps/agreements for changing behaviour. Third, all nursing staff members (and preferably also all other professionals, care staff and support staff that come into contact with the patients) should be trained in working according to the method together; new staff members should also receive the training course to ensure a consistent approach by staff. And finally the necessary prerequisites—such as having an “ABC ambassador” on the ward who puts ABC on the agenda of team meetings and who calls staff members who do not use the ABC method to account—should be in place.

Acknowledgements

We would like to thank all trainers and all patients and nurses of SVRZ Ter Poorteweg Koudekerke, Department of Acquired Brain Injury Huize Padua GGZ Oost Brabant, rehabilitation centre Adelante Hoensbroek and Hambos clinic Kerkrade for their participation in this study. We would like to thank Stichting ABC99 for the supply of ABC training materials.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek under [grant number 056-11-013].