Disruptive behaviour and antipsychotic medication use in long term care homes

Abstract

The study used hierarchical linear modeling to derive a profile of disruptive behaviours that predicted the use of antipsychotic medication in long-term care homes. The findings showed significant differences across homes in antipsychotic use that was not attributable to mean levels of disruptive behaviour. At the level of residents, the predictors included verbally abusive, physically abusive, and socially inappropriate behaviour, but not resistance to care.

Introduction

The rate of antipsychotic use with residents of long-term care homes (LTCH) is higher than that for age-matched peers living independently. Studies in Ontario and Quebec showed a mean prescription rates within a 30-34% range compared with a 3% reference rate for the population at large (Gobert & D’Hoore, 2005; Rapoport, Mamdani, Shulman, Herrmann & Rochon, 2005; Rochon, Stukel, Bronskill, Gomes, Sykora, Wodchis, et al. 2007). Although prescribed to counter-disruptive behaviour, both conventional and atypical antipsychotics have serious side effects, with several studies showing elevated rates of mortality after a period of use (Bender, 2006; Falsetti, 2000; Knol, van Marum, Jansen, Souverein, Schobben, & Egberts , 2008). Consequently, these medications are not without risk and countries differ in the measures taken to reduce that risk.

The United States, for example, has stricter criteria for the prescription of antipsychotic medication in LTCH than does Canada. A federal law passed in 1987 (OBRA ’87) to protect residents of nursing homes and assisted-living facilities that receive government funding, restricts the use of antipsychotics to the treatment of agitation, aggression and psychotic behaviour that is distressing to residents or dangerous to others. Moreover, alternative approaches to symptom management failed to resolve the problem (Winzelberg, 2003). There is also an expectation of dosage reduction after six months of medication (Shorr, Fought, & Ray, 1994).

The present research is less concerned with safety issues associated with the use of antipsychotic medication than with shortcomings in previous research on its relationship with disruptive behaviour. The study addresses two methodological problems. The first is that measures of disruptive behaviour in LTCH are not uniform. Voyer, Verreault, Mengue, Laurin, Rochette, Martin and Baillargeon (2005) used 29 items within four categories of aggressive physical behaviour, nonaggressive physical behaviour, aggressive verbal behaviour and non-aggressive verbal behaviour. Burton, Rovner, German, Brant & Clark (1995), in a one-year longitudinal study, found seven behaviours associated with antipsychotic use: wandering, pacing, exhibiting objectionable behaviour, making intolerable noises, interfering with staff, refusing staff instructions and hitting, biting and scratching. Brink (2008) used frequencies of verbally and physically abusive behaviour from the Resident Assessment Instrument 2.0 (RAI 2.0). Stones, Stewart and Kirkpatrick (2003) measured disruptive behaviour with an 8-item scale from the RAI 2.0 that encompassed frequency and alterability of verbal abuse, physical abuse, socially inappropriate behaviour, and resistance to care. Although the scoring of the RAI 2.0 is specific about the symptoms falling with the aforementioned domains, no previous study separated out domainspecific relationships with antipsychotic use nor differentiated between the frequency and alterability of behaviour within those domains. Because of the widespread use of RAI 2.0 in LTCH in Canada and other countries, one aim of the present study was to clarify multivariate relationships between the items measuring disruptive behaviour with antipsychotic use.

A second aim was to rectify a problem that occurs because of assumptions common to statistical analyses in previous research. These assumptions are that residents from different LTCH belong to the same random sample and are independent of each other. Unfortunately, these assumptions are patently false given abundant evidence that residents do interact with each other, influence each other behaviourally, and that different homes vary in their treatment of residents. A way to resolve this problem is with hierarchical linear modeling, wherein LTCH is designated a random variable with residents nested within their respective LTCH.

Methods

Participants

Data on antipsychotic use was availabe for 1,541 residents of 21 LTCH in Ontario from the RAI 2.0 collected as part of the Resident Assessment Instrument Health Infomatics Project in 2000-2001. Analyzed data was from the first phase of collection.

Materials

The outcome measure was the presence or absence of daily antipsychotic use during the seven days before assessment. The predictor measures were verbally disruptive behaviour, physically disruptive behaviour, socially inappropriate behaviour, and care resistance. Each of eight items assessed either the frequency and alterability of behaviour within one of four the respective domains of disruptive behaviour during the seven days before assessment. Assessment of frequency used a four-point scale depending on whether the behaviour was not present, present on days 1-3, present on days 4-6, or was a daily occurrence. Assessment of alterability used a binary scale indicating that the behaviour was absent or easily altered versus behaviour that was present and difficult to alter.

Data Analysis

The program used for data analysis was version six of Hierarchical Linear Modeling (HLM 6). Because of nesting of residents within LTCH, the model assumed the latter to be a random variable and used a logit link to predict daily antipsychotic use in separate analyses for the frequency and alterability items. The predictor variables included four domains of disruptive behaviour at the resident level (i.e., Verbal1, Physical1, Inappropriate1, and Resistive1 behaviour) and the LTCH means for those domains (i.e., Verbal2, Physical2, Inappropriate2, and Resistive2). Its specification is as follows:

Prob(ANTIPSYCHOTIC=1|B) = P, where log[P/(1-P)] = B0 + B1*(VERBAL1) + B2*(PHYSICAL1) + B3*(INAPPROPRIATE1) + B4*(RESIST1) + G01*(VERBAL2) + G02*(PHYSICAL2) + G03*(INAPPROPRIATE2) + G04*(RESIST2) + U0, with resident level predictors centred on LTCH means and LTCH level predictors centred on grand means. The parameter estimates were from a population average model.

Results

Preliminary analyses showed that all but 24 of the 1,541 residents received antipsychotic medication either daily or not at all during the seven days before assessment. The number of residents with medication was 373 and 1,168 received no medication. The proportion of residents receiving antipsychotic medication within all surveyed LTCH ranged from 10% - 47.1%.

Total proportions of residents with disruptive behaviour ranged from 19.3% for socially inappropriate behaviour to 45% showing resistance to care. Total proportions of residents with disruptive behaviour that was hard to alter ranged from 15.4% for physically abusive behaviour to 32.5% for resistance to care.

Parameter estimates from the HLM analysis showed a significant random effect for LTCH with both the frequency (X 2[17] = 46.51, p < .001) and alterability items (X 2[17] = 48.67, p < .001). Fixed effects were significant at the resident level for verbally abusive, physically abusive, and socially inappropriate behaviour with respect to both frequency and alterability. Fixed effects were nonsignificant at the LTCH level. Table 1 and Table 2 show the coefficients and related statistics.

Further analyses included interactions between resident level and LTCH level measures within each disruptive behaviour domain. None of the interactions was significant.

Discussion

Although the data analyzed in this study were below census level, the findings show significant differences between LTCH in rates of prescribing antipsychotic medication. Given that the LTCH means for each of the disruptive behaviour measures failed to predict antipsychotic use, the differences in prescribing practice is not attributable to mean levels of disruptive behaviour across homes. However, at the resident level, verbally abusive, physically abusive, and socially inappropriate behaviour were all predictors of antipsychotic use whether assessed by either frequency or alterability. Resisting care was not a significant predictor. These resident level findings illustrate a potential pitfall of integrating different (albeit correlated) disruptive behaviours within a single scale, with analysis of the multivariate profile showing differences in relationship with the predicted variable.

An implication of the findings is that verbal abuse, physical abuse, and socially inappropriate behaviour are reasons for prescribing antipsychotic medication. These are all behaviours noted in residents with dementia or other neurological or psychiatric impairments. Brink (2008) found compatible evidence that psychiatric diagnosis, cognitive impairment, verbal abuse, and functional limitations were predictive of antipsychotic use in LTCH.

Another implication derives from findings of differences among LTCH homes. Although LTCH homes differed in rates of prescribing antipsychotics, these differences were not attributable to mean levels of disruptive behaviour. We are unable to explain them with the measures analyzed in the present study. Whether they related to mean levels on unanalyzed measures (e.g., cognitive impairment, psychiatric diagnosis, etc.) or the effectiveness of alternative intervention to manage disruptive behaviour are questions for future research.

This article will conclude with brief mention of alternative approaches to the management of disruptive behaviour. Research previously reported in this journal found that disruptive behaviour in LTCH residents related to stress, discomfort, untreated depression, dental pain, and delirium (Stones, Stewart, & Kirkpatrick, 2003). Studies reported elsewhere showed that non-pharmaceutical treatment of disruptive behaviour to be effective given requisite levels of staffing and training (Dwyer-Moore & Dixon, 2007; James, Mackenzie, Pakrasi, & Fossey, 2008; Plaud, Moberg, & Ferraro, 1998). The time may yet come when antipsychotic use becomes the treatment of last resort.