Terapia de presencia simulada para la demencia
Resumen
Antecedentes
La demencia es un síndrome neuropsiquiátrico común y grave, caracterizado por deterioro cognitivo y funcional progresivo. La mayoría de los pacientes con demencia desarrolla trastornos conductuales, también conocidos como síntomas conductuales y psicológicos de la demencia (SCPD). Se han evaluado varias intervenciones no farmacológicas para tratar los SCPD en los pacientes con demencia. La terapia de presencia simulada (TPS), una intervención que utiliza grabaciones en video o cintas de audio de los miembros de la familia reproducidos a la persona con demencia, es un enfoque posible para tratar los SCPD.
Objetivos
Evaluar los efectos de la TPS sobre los síntomas psicológicos y conductuales y la calidad de vida en los pacientes con demencia.
Métodos de búsqueda
Se hicieron búsquedas en ALOIS (registro especializado del Grupo Cochrane de Demencia y Trastornos Cognitivos [Specialised Register of the Cochrane Dementia and Cognitive Improvement Group]), CENTRAL (The Cochrane Library) (9 abril del 2020), MEDLINE Ovid SP (1946 hasta 9 de abril del 2020), Embase Ovid SP (1972 hasta 9 de abril del 2020), PsycINFO Ovid SP (1806 hasta 9 de abril del 2020), CINAHL vía EBSCOhost (1980 hasta 9 de abril del 2020), LILACS vía BIREME (todas las fechas hasta 9 de abril del 2020), ClinicalTrials.gov (ClinicalTrials.gov) (todas las fechas hasta 9 de abril del 2020), y en el World Health Organization (OMS) Portal (apps.who.int/trialsearch) (todas las fechas hasta 9 de abril del 2020). También se revisaron las listas de referencias de artículos relevantes para identificar estudios adicionales.
Criterios de selección
Ensayos controlados aleatorizados y cuasialeatorizados, incluidos los estudios cruzados (cross‐over), que evaluaran la eficacia de la TPS, que consta de grabaciones personalizadas de audio o video de los miembros de la familia, en pacientes con cualquier forma de demencia.
Obtención y análisis de los datos
Dos revisores seleccionaron los estudios de forma independiente, evaluaron el riesgo de sesgo y extrajeron los datos. No se realizó ningún metanálisis debido a la heterogeneidad apreciable entre los estudios incluidos.
Resultados principales
Tres ensayos con 144 participantes cumplieron los criterios de inclusión. Dos de los ensayos tuvieron un diseño aleatorizado cruzado, uno fue un ensayo cruzado que se clasificó como cuasialeatorizado.
Los participantes de los estudios incluidos eran pacientes con demencia que vivían en centros de atención. Eran predominantemente mujeres y tenían una media de edad de más de 80 años. Se realizó la TPS mediante una grabación de audio o video preparada por miembros de la familia o substitutos. Varió en cuanto al contenido, la frecuencia de administración y la duración. Todos los estudios compararon múltiples tratamientos. En un estudio, la TPS se comparó con otras dos intervenciones. En los otros dos estudios, se comparó con otras tres intervenciones. Específicamente, la TPS se comparó con atención habitual, música personalizada (dos estudios), una cinta de audio de “placebo” que contenía la voz de una persona (dos estudios), e interacción social individual realizada por asistentes de investigación adiestrados (un estudio). En lo que respecta a los resultados evaluados, un estudio consideró la agitación y el retraimiento (ambos evaluados con tres métodos). El segundo estudio evaluó la conducta disruptiva verbal (evaluada con tres métodos); y el tercer estudio evaluó la conducta físicamente agitada y la conducta verbalmente agitada (el método utilizado no se describió claramente).
Según los criterios de GRADE, la calidad general de la evidencia fue muy baja debido al número muy reducido de participantes y al riesgo de sesgo en los estudios incluidos; (ninguno de los ensayos presentó un bajo riesgo de sesgo de selección; todos los ensayos tuvieron un alto riesgo de sesgo de realización; un ensayo tuvo un alto riesgo de sesgo de desgaste; y todos tuvieron una información selectiva poco clara).
Debido a la variación en los participantes, el formato de la TPS, las intervenciones de comparación y las medidas usadas para evaluar los resultados, los hallazgos se consideraron no apropiados para un metanálisis.
En cada ensayo, el efecto de la TPS sobre la conducta, comparado con la atención habitual, fue contradictorio y dependió de la medida utilizada. Dos ensayos que incluían una intervención con música personalizada no informaron ninguna diferencia significativa entre la presencia simulada y la música en los resultados conductuales. Debido a que la calidad general de la evidencia fue muy baja, no hubo mucha seguridad con respecto a todos los resultados
Ninguno de los estudios evaluó la calidad de vida ni alguna de las medidas de resultado secundarias (realización de las actividades cotidianas, abandonos y carga para el cuidador).
Conclusiones de los autores
No fue posible establecer conclusiones acerca de la eficacia de la TPS para el tratamiento de los síntomas conductuales y psicológicos y para mejorar la calidad de vida de los pacientes con demencia. Se necesitan nuevos estudios de alta calidad para investigar el efecto de la TPS.
PICO
Resumen en términos sencillos
Terapia de presencia simulada para la demencia
Pregunta de la revisión
¿La terapia de presencia simulada (TPS) puede tratar las conductas problemáticas y mejorar la calidad de vida en los pacientes con demencia?
Antecedentes
La demencia es una enfermedad, más comúnmente observada entre personas mayores, en la cual la memoria y otras funciones cerebrales se deterioran y los pacientes se vuelven gradualmente dependientes de otros para la atención. Muchos pacientes con demencia, en particular en los estadios posteriores, a veces manifiestan signos de dificultad, o se comportan de maneras difíciles de manejar para los cuidadores. Hay mucho interés en encontrar formas de tratar este trastorno sin utilizar fármacos. La terapia de presencia simulada es un tipo de tratamiento que se ha usado principalmente en residencias. Incluye la reproducción de una grabación de video o una cinta de audio personalizada de los miembros de la familia a la persona con demencia cuando presenta angustia o agitación.
Características de los estudios
Se realizaron búsquedas de los ensayos que comparaban la TPS con atención habitual u otro tratamiento. En condiciones ideales, los pacientes con demencia debían haber sido asignados al azar a uno u otro tratamiento, aunque también se incluyeron ensayos aunque la asignación al tratamiento no hubiese sido estrictamente aleatoria.
Se encontraron tres ensayos que cumplieran los criterios de inclusión. Los 144 participantes estaban viviendo en centros de atención. La mayoría eran mujeres con una edad promedio de más de 80 años y demencia grave. La forma de administrar la TPS fue diferente en cada ensayo. Todos los ensayos usaron más de un tratamiento de comparación, que difirió entre los ensayos. Todos los ensayos intentaron medir un efecto sobre las conductas de agitación, aunque utilizaron enfoques diferentes.
Hallazgos clave
Debido a que los ensayos fueron tan diferentes entre sí, no fue posible agrupar los resultados. Individualmente, cada ensayo informó diferentes métodos para evaluar el efecto de la TPS sobre los problemas conductuales y los resultados variaron según el método utilizado para medir el resultado.
Ninguno de los estudios evaluó la calidad de vida, el efecto sobre las actividades diarias, los efectos sobre los cuidadores, ni las probabilidades de que los participantes abandonaran el estudio.
Calidad de la evidencia
Los estudios fueron pequeños y todos tenían problemas con los métodos, lo cual podría haber sesgado los resultados. En consecuencia, se cree que la calidad general de la evidencia fue muy baja, lo cual significa que no es posible confiar del todo en los resultados.
Conclusión
No se ha realizado suficiente investigación de alta calidad que permita juzgar si la TPS puede ayudar a los pacientes con demencia que presentan angustia o agitación.
Authors' conclusions
Summary of findings
| Simulated presence therapy compared with usual care for people with dementia | |||
| Patient or population: dementia | |||
| Outcomes | Impact | № of participants | Quality of the evidence |
|---|---|---|---|
| Overall BPSD (multidomain) – not assessed | ‐ | ‐ | ‐ |
| Agitation | One study reported that agitation assessed by two methods (direct observation and weekly staff survey) was not reduced by SPT, while a third method (staff observation logs) showed a decline in agitation. In a second study, two methods (VDB score and standardised observational data) showed that SPT reduced verbally disruptive behaviour, while there was no effect on the same outcome with a third method (nurses’ assessments). A third study with an unclear assessment method reported that SPT was effective in reducing physical agitation and verbal agitation compared to usual care. | 144 | ⊕⊝⊝⊝ |
| Depression – assessed in one study as "withdrawn behaviours" which included a lack of interest and sad mood | SPT did not improve withdrawn behaviours, interest or mood based on direct observations. Weekly staff surveys did not show any effect of SPT on mood, although the intervention was associated with increased interest. The staff, through observation logs, reported that SPT compared to usual care statistically significantly improved withdrawn behaviour. | 54 (1 trial) | ⊕⊝⊝⊝ |
| Anxiety ‐ not measured | ‐ | ‐ | ‐ |
| Quality of life ‐ not measured | ‐ | ‐ | ‐ |
| Dropouts ‐ not measured | ‐ | ‐ | ‐ |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||
| GRADE Working Group grades of evidence | |||
| 1 Downgraded by two levels because of very serious concern about the risk of bias: none of the trials was at low risk of selection bias (one study was at high risk of bias); all the trials were at high risk of performance bias; one trial was at high risk of attrition bias; all had unclear selective reporting; and two were at high risk of other biases. 2 Downgraded by two levels because of serious concern about imprecision: none of the studies reached the optimal information size. 3 Downgraded by two levels because of serious concern about the risk of bias; the trial was at unclear risk of selection bias; high risk of performance bias; unclear risk of selective reporting bias; and high risk of other bias. 4 Downgraded by two levels because of very serious concern about imprecision: very small sample size. | |||
Background
Description of the condition
Dementia is a common and serious neuropsychiatric syndrome, characterised by progressive cognitive and functional decline. The prevalence of dementia is estimated to be 6% in populations aged 60 years or older (Prince 2013). It increases dramatically with age, reaching 43% in people 85 years or older (Thies 2012).
Dementia imposes a considerable burden on families and is becoming a major socioeconomic challenge for the healthcare system (Wimo 2013). It is one of the strongest predictors of nursing home admission and is a major clinical issue for nursing home residents. Its prevalence varies from 26% to 48% in US nursing homes to over 60% in European nursing homes (Cherubini 2012).
The majority of people with dementia develop behavioural disturbances, also known as behavioural and psychological symptoms of dementia (BPSD), over the course of the condition. BPSD comprise a heterogeneous group of non‐cognitive symptoms, such as agitation, disruptive behaviour, aberrant motor behaviour including wandering, irritability, hallucinations, anxiety, depression, apathy, delusions, disinhibition, and appetite and sleep alterations (Cerejeira 2012). BPSD are a major component of the dementia syndrome as they are strongly correlated with the severity of functional impairment, occurring most frequently in people with dementia in hospitals and nursing homes (Cerejeira 2012). These symptoms are independently associated with poor outcomes such as distress among those with dementia and their caregivers, inappropriate prescribing of antipsychotic medication, long‐term hospitalisation, mortality, excess morbidity, institutionalisation, early placement in a nursing home, and increased healthcare costs (Borson 1997; Clyburn 2000; Kales 2015; Van Den Wijngaart 2007).
Several studies have documented that almost all people with dementia experience at least one episode of behavioural disturbance at some point during their illness (Kunik 2010; Lyketsos 2007; Spalletta 2010; Steinberg 2008). In a cohort of 408 dementia participants, for example, the Cache County Study estimated the five‐year period prevalence of neuropsychiatric symptoms and found that 97% of the participants experienced at least one symptom. Five‐year period prevalence was highest for depression (77%), apathy (71%), and anxiety (62%). Rates for agitation/aggression vary between 13% and 24% (Steinberg 2008). Behavioural symptoms frequently occur together (for example wandering with sleep problems or irritability with persecutory delusions) (Savva 2009).
Description of the intervention
Behavioural problems in people with dementia are often treated with drugs, which may lead to several adverse outcomes including sedation, falls, extrapyramidal disturbances, stroke, and increased mortality (Lopez 2013; Sink 2005; Stern 1994). For this reason, non‐pharmacological interventions are an appealing alternative (Abraha 2015b); and have been recommended as a first choice of treatment for BPSD (Salzman 2008). Several non‐pharmacological interventions have been evaluated to treat BPSD in people with dementia (Abraha 2017). These include emotion‐oriented therapies (such as reminiscence therapy (Woods 2009), simulated presence therapy (Zetteler 2008), validation therapy (Neal 2003)), sensory stimulation interventions (such as acupuncture (Peng 2007), aromatherapy (Forrester 2014), light therapy (Forbes 2014), massage/touch therapy (Viggo Hansen 2006), music therapy (Vink 2004), Snoezelen multisensory stimulation (Chung 2002), and transcutaneous electrical nerve stimulation (Cameron 2003)), and other interventions (such as functional analysis‐based interventions (Moniz‐Cook 2012), and exercise therapy (Forbes 2013)).
It has been reported that psychosocial interventions work best to reduce agitation and improve other behavioural disturbances in people with dementia when they are personalised to reflect the individual's background and environmental circumstances (O'Connor 2009).
The focus of this review is simulated presence therapy (SPT), in which video or audiotape recordings of family members are played to the person with dementia (Woods 1995). The content of the recordings may vary depending upon the interests of the individual and may include conversations, stories, or shared memories. The intention is that the recorded voice of a family member is reassuring and that anxiety and distress are reduced by making the environment of the person with dementia as familiar as possible (Woods 1995).
How the intervention might work
SPT is sometimes referred to as simulated family presence therapy (Garland 2007), because it originated from the observation that nursing home residents who received more visits from family members were less agitated and had greater life satisfaction (Woods 1995). Since family members may not be able to visit frequently, and family visits may be forgotten rapidly by residents with dementia, SPT aims to reduce the separation anxiety experienced by the resident (Miesen 1993; Peak 2002).
The technique was first described in a 1995 study by Woods and Ashley, in which 27 nursing home residents with dementia listened to a tape prepared by their caregivers. The authors reported a substantial reduction of behavioural problems such as verbal aggression and social isolation, and an increase in positive behaviours such as proper verbalisation, smiling, and singing (Woods 1995). In the following years, a limited number of studies have been performed to evaluate the efficacy of SPT. These studies are characterised by considerable methodological heterogeneity, including different outcomes, as some studies examined the effect of SPT on quality of life, while other investigations attempted to reduce challenging behaviours.
Although SPT has been investigated for more than 20 years, the mechanisms by which SPT might work are still unknown. Several explanatory models have been proposed for behavioural disturbances. According to the need‐driven, dementia‐compromised behaviour model, behavioural disturbances of the cognitively impaired person are an expression of an unmet need or goal (Algase 1996). These disturbances might therefore be seen as an expression of the person seeking help or contact from family members. In this respect SPT might fulfil this unmet or personal need.
Another conceptual framework, the progressively lowered stress threshold model, proposes that people with dementia become progressively unable to understand their environment (for example recognize objects and their purposes and use). As a consequence, people with dementia experience stress that causes a variety of behavioural disturbances (Smith 2004). In this context, voices and images of family members may represent well‐known references that help to lower stress levels.
Generally, the literature is scarce in terms of the epidemiology of adverse events associated with non‐pharmacological interventions. In the case of SPT, adverse events cannot be excluded, including the possibility that the intervention might worsen agitation or mood. Woods and Ashley reported that in 7% of the observation during SPT treatment, behaviours either remained unchanged or worsened with the audiotape (Woods 1995). In some cases, for example, material contained in the tapes could be distressing for the person with dementia (Peak 2002; Garland 2007).
Why it is important to do this review
The majority of BPSD, in particular agitation, disruptive behaviour, and aberrant motor behaviours, are currently treated with psychotropic drugs that are not very effective (Howard 2012; Katz 2007; Olin 2002), have several important adverse effects including increased mortality (Budnitz 2011; Opie 1999), and result in excessive drug costs and healthcare utilisation (Leendertse 2011; Rottenkolber 2011). Non‐pharmacological interventions should therefore be the first choice to treat BPSD.
SPT is a simple intervention that does not require expensive equipment, nor does it require extensive training. In addition, SPT reproduces an encounter of the patient with a family member, which is usually a positive experience for the former. The potentially relevant adverse event is an increase in agitation, which should not cause a serious risk for the patient's health. As such, we believe that it is important to verify whether SPT is effective in treating BPSD in people with dementia.
Objectives
To assess the effects of SPT on behavioural and psychological symptoms and quality of life in people with dementia.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised and quasi‐randomised controlled trials, including cross‐over studies, of SPT in people with dementia.
Types of participants
People with any form and severity of dementia diagnosed according to the International Classification of Diseases, Revision 10 (ICD‐10), Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM‐IV), DSM‐5, or comparable criteria. No restriction in terms of setting was applied.
Types of interventions
Active treatment
SPT consisting of audio or videotape recording that family members or caregivers have personalised. The content of the recording should include positive experience from the participant's past life and shared memories involving family or close friends.
Comparison
-
Usual care with no additional activity.
-
Any activity that differs in content and approach from SPT, but is additional to usual care (for example, music therapy, reminiscence groups, reality orientation groups, or social contact groups that do not use the techniques identified as SPT).
Types of outcome measures
Primary outcomes
-
Any behavioural and psychological symptoms as measured using scales such as the Neuropsychiatric Inventory (NPI) or Brief Psychiatric Rating Scale (BPRS) (multidomain scales), Cohen‐Mansfield Agitation Inventory (CMAI) (scale specific to agitation), or Cornell Scale for Depression in Dementia (CSDD) (scale specific to depression).
-
Quality of life.
Secondary outcomes
-
Activities of daily living.
-
Caregiver burden.
-
Dropouts (as a measure of acceptability).
Search methods for identification of studies
Electronic searches
We searched ALOIS (www.medicine.ox.ac.uk/alois), the Cochrane Dementia and Cognitive Improvement Group’s Specialised Register (up to 9 April 2020).
ALOIS is maintained by the Information Specialists for the Cochrane Dementia and Cognitive Improvement Group and contains studies that fall within the areas of dementia prevention, dementia treatment and management, and cognitive enhancement in healthy elderly populations. The studies are identified through:
-
monthly searches of a number of major healthcare databases: MEDLINE, Embase, CINAHL, PsycINFO, and LILACS;
-
monthly searches of a number of trial registers: ISRCTN, UMIN (Japan's trial register), the World Health Organization International Clinical Trials Registry Platform (which covers ClinicalTrials.gov, ISRCTN, the Chinese Clinical Trials Register, the German Clinical Trials Register, the Iranian Registry of Clinical Trials, and the Netherlands National Trials Register, plus others);
-
quarterly searches of the Cochrane Library’s Central Register of Controlled Trials (CENTRAL);
-
six‐monthly searches of a number of grey literature sources: ISI Web of Knowledge Conference Proceedings, Index to Theses, Australasian Digital Theses.
To view a list of all sources searched for ALOIS see About ALOIS on the ALOIS website (www.medicine.ox.ac.uk/alois).
Details of the search strategies run in healthcare bibliographic databases, used for the retrieval of reports of dementia, cognitive improvement, and cognitive enhancement trials, can be viewed in the ‘Methods used in reviews’ section within the editorial information about the Cochrane Dementia and Cognitive Improvement Group.
We ran additional searches in MEDLINE (Ovid SP), Embase (Ovid SP), PsycINFO (Ovid SP), CINAHL (EBSCOhost), ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform to ensure that the searches for the review were as comprehensive and up to date as possible. We have presented the search strategies used in Appendix 1.
We carried out translations for non‐English papers, where possible.
Searching other resources
We searched the grey literature, such as conference proceedings, through Embase (Ovid SP) and through Web of Science (ISI Web of Science), and the reference lists of all the potentially relevant trials or reviews identified through the above searches. We applied no language restriction.
Data collection and analysis
Selection of studies
In order to facilitate the retrieval of articles of interest, we downloaded all references into a single database using the EndNote reference management software (Endnote 2013).
We obtained full‐text copies that two review authors independently assessed for eligibility (IA, JMR). The review authors resolved any disagreements by discussion.
Data extraction and management
For eligible studies, pairs of review authors (JMR, IA and IL, MV) independently extracted the data using the Cochrane form available in Review Manager 5 (RevMan 5) software (Review Manager 2014). They resolved disagreements by discussion.
We extracted the following information from each included study.
-
General information about the study (year of publication, setting).
-
Study design, methods of recruitment, inclusion and exclusion criteria, details of the control and comparison groups, and incentives for participation.
-
Characteristics and number of participants.
-
Characteristics of the intervention.
-
Results.
One review author (IA) entered data into the Review Manager 5 software (Review Manager 2014), and a second review author checked the data for accuracy.
Assessment of risk of bias in included studies
We used Cochrane's recommended tool for assessing risk of bias (Higgins 2011).
We assessed the following domains of each included study: sequence generation and allocation concealment for selection bias (Savovic 2012; Wood 2008); blinding of participants or personnel for performance bias (Savovic 2012; Wood 2008); blinding of outcome assessors for detection bias (Savovic 2012; Wood 2008); incomplete outcome data for attrition bias (Abraha 2015a); selective reporting bias (Chan 2004; Macura 2010); and other potential sources of bias (Higgins 2011). Other sources of bias included the comparability of intervention and control group characteristics at baseline, validation of outcome assessment tools, and reliability of outcome measures.
Review authors (IA, JMR, ILM, MV) independently evaluated the risk of bias of the included studies, using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). They resolved any disagreement by discussion; or in the event of deadlock, by consulting with a third review author.
We described the quality assessment in a 'Risk of bias' table.
Measures of treatment effect
Where a sufficient number of trials with homogeneous populations were identified, we planned to carry out meta‐analyses of primary and secondary outcomes. For the analysis of dichotomous outcomes, we planned to use risk ratios with 95% confidence intervals (CI); for the analysis of continuous outcomes, we planned to use the mean difference or standardised mean difference and 95% CI.
Unit of analysis issues
The unit of analysis was the individual participant with dementia.
In case of studies that contributed with multiple, correlated comparisons, we planned to combine study groups into a single pair‐wise comparison.
In the event of cluster‐randomised trials, if the reported analysis did not correctly account for the cluster design we planned to consider whether it was possible to estimate the effective sample size using the intracluster correlation coefficient. If this was not possible, then we planned to conduct sensitivity analyses excluding the trials that did not originally adjust for clustering.
Dealing with missing data
In the event of missing data, we contacted the corresponding authors to try to obtain the relevant information.
Where 'intention to treat' was not available, we planned to conduct analyses using 'available cases' data (i.e. only those whose results are known).
Assessment of heterogeneity
We planned to assess heterogeneity of study characteristics and statistical heterogeneity. We evaluated heterogeneity by examining the relative table of characteristics of the included population, the type of interventions, and the type of outcome measures.
Where meta‐analysis could be performed, we planned to appraise statistical heterogeneity through a visual assessment of the forest plot, in addition to evaluating the Chi² test and I² statistic. We planned to consider a Chi² test with a P value of 0.10 to be significant, and we planned to interpret the I² statistic as: 0% to 40% indicates unimportant heterogeneity; 30% to 60% indicates moderate heterogeneity; 50% to 90% indicates substantial heterogeneity; and 75% to 100% indicates considerable heterogeneity (Deeks 2011).
Assessment of reporting biases
We planned to use a funnel plot to examine for small‐study effects, which may indicate publication bias, had there been more than 10 studies that could be combined into a meta‐analysis,
Data synthesis
We planned to pool data for analysis if the interventions, outcomes, and participant populations in the individual studies had been sufficiently similar (to be determined by consensus).
We planned to use Review Manager 5 for all analysis (Review Manager 2014).
Subgroup analysis and investigation of heterogeneity
We planned subgroup analyses based on several participant characteristics such as sex, severity of dementia, severity of behavioural problems at baseline, the nature of the control intervention (treatment ‘as usual’ or another active treatment), stage of dementia, care setting (community versus long‐term care facility), and intervention characteristics such as length of treatment, including number of sessions, and mode of delivery (audio versus video recording).
Sensitivity analysis
We planned to perform sensitivity analyses to assess the consistency and robustness of the results of the meta‐analysis based on:
-
including only studies with low risk of selection bias (sequence generation and allocation concealment);
-
including only trials with low risk of detection bias (blinding of outcome assessor);
-
including only trials with low risk of attrition of bias;
-
including only trials that used an intention‐to‐treat analysis;
-
repeating analyses using a fixed‐effect model for data synthesis.
Summary of findings table
We used the GRADE approach to assess the quality of the supporting evidence behind each estimate of treatment effect for the comparison of SPT with usual care. We used risk of bias, imprecision, inconsistency, indirectness, and publication bias to rate the overall evidence (Balshem 2011).
We present key findings of the review in 'summary of findings Table 1'. This includes a summary of the amount of data and the overall quality of the evidence for overall behavioural and psychological symptoms, agitation, depression, anxiety, quality of life and dropouts (Schünemann 2011a; Schünemann 2011b).
Results
Description of studies
Results of the search
A systematic search of electronic databases combined with a handsearch of references identified 299 publications (Figure 1). After removing duplicate records, 229 records were screened of which 216 were removed and 13 were evaluated in full‐text. Ten of these, which were either not trials or were outside the scope of this review, were excluded. Finally, three trials remained for inclusion (Camberg 1999; Cohen‐Mansfield 1997; Garland 2007).
Study flow diagram.
Included studies
The three included trials initially enrolled 144 participants of whom 116 were evaluated by the authors in the final analysis (Cohen‐Mansfield 1997 initiated with 60 participants and evaluated only 32). For full details, see the Characteristics of included studies table.
Setting
Two trials were conducted in the USA (Camberg 1999; Cohen‐Mansfield 1997); and one in Australia (Garland 2007).
All of the studies were conducted in a nursing home setting.
Study design
Two trials were randomised cross‐over trials (Camberg 1999; Garland 2007); and one was a cross‐over trial which we classified as quasi‐randomised (Cohen‐Mansfield 1997). In this trial, participants were assigned to four groups, each of which rotated through the active treatments (n = 3) and control treatment (n = 1) in a different order so that by the end of the evaluation period all groups were exposed to all the treatments and to the no‐treatment period (Cohen‐Mansfield 1997).
Participants
Camberg 1999 included 54 participants of whom 48 (89%) were women. The average age was 82.7 years (standard deviation (SD) 7.5 years). The participants had severe cognitive impairment (Mini Mental State Examination (MMSE): 5.1, SD ± 4.4; Texas Success Initiative (TSI) Assessment: 13.1 SD ± 6.9) and low functional ability (Bedford Alzheimer's Nursing Scale (BANS): 13.2, SD ± 4.0; Activity of daily living (ADL) Self‐Performance Scale: 2.7, SD ± 1.2). No other baseline information was provided. Participants were selected based on the presence of at least one agitated behaviour per day of those listed on the short form of the Cohen‐Mansfield Agitation Inventory Scale or one indication of withdrawn behaviour “defined either by "sounding sad" or "crying", that occurred at least "often"; or "seldom" interested in activities, social interaction, or in their immediate environment, from the Multidimensional Observation Scale for Elderly Subjects (MOSES)”
At baseline, the authors determined the prevalence of agitated behaviours from 6903 direct observations and 222 weekly staff surveys. The most frequent kinds of physical agitation were “repetitive motions in place, outward motions with the extremities, and pacing/aimless wandering”, while for verbal agitation the most recurrent types were “repetitive vocalization, cursing/verbal aggression, and negative words or attitude”. Most agitated behaviours were quite mild and only occurred for a few minutes; and in 25% of direct observations, agitation was not observed at all. The authors also measured the prevalence of withdrawn behaviours as interest and mood. Subjects showed no interest in almost half (47%) of the direct observations and in more than half (55% to 72% based on the category of interest) of the weekly staff surveys. Regarding mood, subjects displayed no pleasure in 75% of direct observations and in nearly 40% of weekly staff surveys. During most (82.5%) of the direct observations, subjects were reported to have “no expression” on their faces.
In Cohen‐Mansfield 1997, the inclusion criterion was the presence of verbally disruptive behaviours (VDBs) manifested at least several times a day. In this study, 60 participants in a nursing home were identified and provided consent. Of these, five died before any data were collected, 11 became quiet, three refused to continue, and two were physically restrained and thus excluded. Hence, 32 participants remained with complete data. These participants had an average age of 86.8 years (SD ± 1.2 years) and 26 were female. At baseline, they had severe functional and cognitive impairment, as shown by the Lawton and Brody Physical Self Maintenance Scale (Activities of Daily Living (ADL)) score (4.0, SD ± 0.2), and the Brief Cognitive Rating Scale score (5.5, SD ± 0.2).
Baseline evaluation in the 32 subjects who completed the study revealed that 11 participants exhibited VDB several times each hour, 18 subjects several times per day and three participants once or twice per day. An analysis of 11,520 3‐minute observations, using the Screaming Behavioural Mapping Instrument (SBMI), before, during and after interventions, revealed that the most frequent VDBs, in order of prevalence, were complaining, shouting, nonsense talk, repeating words and other VDB.
Before initiation of the interventions, participants underwent a medical examination to determine if pain might account for VDB: no physiological signs of pain were found for 23 subjects. Although nine participants received pain medication, physicians stated that the VDB was not caused by pain for any of the subjects.
In Garland 2007, the participants were 30 nursing home residents with a diagnosis of dementia, who had been resident for at least three months and had had at least one disruptive behaviour several times a day in the preceding two weeks (assessed with the CMAI). Mean age was 79 years (range 66 to 93 years) and 63% were female. All participants were prescribed some type of psychotropic medication and had a very low baseline mean MMSE of 2.5 (range 0 to 12). The authors measured pre‐intervention agitated behaviours (2880 observed target behaviours), which consisted of physically aggressive behaviours (3.8%), physically non‐aggressive behaviours (64.8%) and verbally aggressive and non‐aggressive behaviours (31.4%).
Intervention
Camberg 1999 was a three‐intervention study. The authors reported “A Latin Squares crossover design was adopted for this study”.
Each treatment continued for 17 days over 4 weeks, with a 10‐day washout period following each treatment.
The interventions were as follows.
(a) Simulated presence therapy based on a personalised audio tape (SimPres) consisting of a one‐sided telephone conversation with a family member or surrogate, containing many special memories and positive emotions. Family members received a memory inventory form, conversational guidelines, and an instructional audiotape as training material. The study staff coached family members either by telephone or in person, immediately before the recording session. A headset and an auto‐reverse tape recorder were used to play the audio tape;
(b) A placebo, which consisted of a tape made to resemble the SimPres treatment by containing the voice of a person who read emotionally neutral newspaper articles;
(c) Usual care, which was the regular approach used by nursing staff to manage behaviour (e.g. staff interactions, redirection, or physical restraints).
The nursing staff in the study units applied the intervention following a 30‐minute training session during which they were instructed to use the SimPres or placebo audio tapes at least twice a day, Monday to Friday, as an alternative to their usual care, whenever a study subject displayed an agitated or withdrawn behaviour. The nurses were requested not to listen to the audiotapes, but were informed that both tapes contained verbal messages. Procedures and demonstrations of tape use and case examples of the intervention were provided. Adherence to the protocol was ensured, and feedback was given to the study staff by a study monitor, who spent 20 hours per week at each facility.
In Cohen‐Mansfield 1997, participants were assigned to four groups, each of which was exposed to all three interventions and to a control no‐intervention phase. In each group, the order of interventions was different. By the end of the study, all participants were exposed to all types of intervention.
Each group was exposed to each intervention for two consecutive weeks followed by one week of washout.
The interventions were as follows.
(a) A SPT intervention consisting of a videotape, which the relatives of all participants were asked to prepare, was provided to the resident. A research assistant, a nursing assistant, and a volunteer prepared the videotapes for four participants whose relatives either refused or were unable to create a videotape. Family members received general guidelines to prepare the videotapes, but they chose the specific content and the style of communication.
(b) A 30‐minute audiotape recording of music, based on the participant's musical preferences (provided by relatives).
(c) One‐to‐one social interaction performed by trained research assistants. A manual was prepared to guide the social interaction, which included information on how to interact with a nursing home resident, how to start an activity and a broad list of possible activities ordered by the degree of involvement, from high to low, required of the resident. A "box of activities", comprising games, pictures, balls, books, etc. was assembled. Alternatively, the following activities were provided: (i) conversation with a research assistant; (ii) motion exercise (e.g. tossing a foam ball, hand and arm movement); (iii) sensory stimulation (a sensory kit including different fabrics, school supplies, health aids, make‐up, spices, soaps, etc. was available to be presented to the residents to stimulate touch, sight, and smell); (iv) manual activities (such as making a collage, completing a simple puzzle, clipping coupons).
Garland 2007 was a four intervention, cross‐over study. Participants were randomly assigned to three groups and sequentially exposed to all four interventions, with a two‐day washout between each one. Interventions were administered once a day, for three days each, during weeks two, three and four.
The interventions were as follows.
(a) "Simulated family presence", consisting of a professionally edited tape recording of a semi‐structured interview of a family member, conducted by a trained psychologist, regarding participants' earlier lives. The script was designed to resemble a telephone conversation regarding special memories, beloved family members, and family episodes.
(b) A music intervention, which consisted of listening to music based on participants’ selections of popular, big band, and Greek and Dutch music.
(c) A placebo intervention, which was a neutral audiotape, based on non‐personal, non‐emotive material (a psychologist read from a gardening book in neutral tones) to control for the effects of receiving attention from researchers and listening to a tape recording.
(d) Usual care, which was not defined, but the authors stated that “usual care observations were recorded for 45 minutes on each of three days in week 1 of the study at the times of peak agitation nominated by nursing staff”.
Outcome
Camberg 1999 evaluated agitated and withdrawn behaviours. Agitation was defined as an experience of an unpleasant state of excitement, observable without subjective interpretation, not strictly triggered by caregiver activities, not related to known physical needs of the patient that can be alleviated, and without motivational intent. Withdrawn behaviours were defined as a "lack of interest in people, activities, or things in the subject's environment, combined with sad mood". Withdrawn behaviour was assessed considering the combination of "interest", and manifestations of mood. Observers used two items from the Philadelphia Geriatric Centre Affect Rating Scale (PARS), "interest" and "pleasure" to evaluate withdrawn behaviour.
Outcomes were measured by (1) direct observation, (2) staff observation logs, and (3) weekly nursing staff surveys.
Direct observations were made for 3 hours and 20 minutes each week, between 9 a.m. and 7 p.m. by trained, blinded, non‐participant observers. Each observation block was 20 minutes long, divided into four 5‐minute observation intervals. Behaviours were assessed according to the schedule, whether or not a specific intervention was implemented. The following scales were used: (1) a seven‐item Observed Agitation Scale (OAS) developed by the researchers; (2) an agitation visual analogue scale (AVAS); (3) two Positive Affect items, "interest" and "pleasure," from the PARS; (4) a withdrawal visual analogue scale (WVAS); and (5) facial diagrams of mood (FACE).
Staff observation logs were based on the nursing staff's impression of a participant's response to each intervention. (The target behaviour, the intervention and its duration, and the subject’s response (improved, became worse, or stayed the same) were all documented daily, from Monday to Friday).
Weekly staff surveys (administered to a nursing staff member who had cared for the subject for at least three days during the preceding week) used two behavioural rating scales: (1) the short form of the Cohen‐Mansfìeld Agitation Inventory (SCMAI) to record staff recall of the frequency of agitated behaviours; and (2) selected, modified items from the Multidimensional Observation Scale for Elderly Subjects (MOSES) to record the subject's mood and interest.
In Cohen‐Mansfield 1997, VDB was the only outcome assessed. It was defined as “verbal or vocal behaviours that were either repetitive, disruptive or inappropriate to the circumstances in which they were manifested”. It was assessed using three different methods: 1) tape recordings; 2) standardised observations; and 3) informant ratings (nurses’ assessments).
1) VDB duration was registered with a tape recorder located near a subject. The tape was coded by a trained research assistant using a computer program specifically developed for coding video and audiotapes. An episode of VDB was considered to be finished if the resident was silent for one full minute. An overall VDB score was calculated by averaging complaining (weight = 1), disruptive vocalisations (weight = 2) and shouting (weight = 4). Tape recordings of VDB were performed for one hour per day (15 minutes before an intervention, 30 minutes during an intervention, and 15 minutes after an intervention), for 10 days for each of the interventions (music, videotape, social interaction, and no‐intervention).
2) The frequency and context of VDB was examined in standardised observations with the SBMI which assesses nine types of VDB (shouting, screaming, or howling; constant requests for attention; repeating words; complaining or inappropriate verbalizations; cursing; verbal aggression; nonsense talk; hallucinations; and other disruptive verbal behaviours (e.g. groaning and singing)). If a VDB was manifested five or more times, a value of 6 (i.e. constant, C) or a value of 7 (i.e. extreme, E) were registered.
3) In informant ratings, nursing staff very familiar with the subjects rated their VDB using the Cohen‐Mansfield Agitation Inventory (CMAI) on a seven‐point scale (1: never manifests the behaviour; 7: manifests the behaviour several times/hour). Information was collected at the end of each intervention. Duration and frequency of appropriate verbal behaviours were also assessed.
In Garland 2007, participants' actual behaviours were observed by experienced, trained researchers with high interrater reliability. The researchers recorded the presence of behavioural disturbances at two‐minute intervals before, during, and after exposure to the audiotapes with total observation periods lasting 45 minutes each. Behaviours were categorised into one of four groups: (i) physically aggressive agitation; (ii) physically non‐aggressive agitation; (iii) verbally aggressive agitation; and (iv) verbally non‐aggressive agitation. It was unclear which scale was used, although it was stated that the CMAI was used in the period before randomisation. The outcome measure was magnitude of change (a fall in physical/verbal agitation), based on differences in mean behaviour counts, before‐during treatments and during‐after treatments.
Quality of life, performance of activities of daily living and caregiver burden were not evaluated in any of the studies.
Excluded studies
Ten studies were excluded after examination of full texts in 2017 (Characteristics of excluded studies). One was a systematic review that evaluated SPT (Zetteler 2008). Two studies used a pre‐ and post‐test quasi‐experimental design to evaluate the efficacy of SPT (Miller 2001; Woods 1995). Two studies were N‐of‐1 trials: one study included one participant (O'Connor 2011); and the second considered four cases (Peak 2002). One study used SPT consisting of presentation of old photographs and favourite music pieces of individuals with dementia, combined in personalised video channels via a television monitor. Participants were 23 people with Alzheimer's disease but insufficient information was provided as to how the participants were allocated between the groups or whether the study was a cross‐over trial (Kajiyama 2007). In one trial, the authors stated that the three interventions (SPT, music and usual care) were offered in an alternating pattern, but also that the six participants were offered a choice of interventions (Cheston 2007). One was an exploratory study that had the goal of collecting information about the delivery of Personal Message cards, as an alternative to simulated presence therapy (Evans 2016). The last study was an abstract of a protocol only (Smith 2010). We added three further studies to the excluded studies list after the updated search in 2020. One was a small qualitative study (Hung 2018), one was a single case study (Lim 2018) and one was an RCT of SPT for hospitalised patients with delirium (Waszynski 2018).
Risk of bias in included studies
The overall quality of the studies was considered low (Characteristics of included studies). Figure 2 shows our judgements about each risk of bias item presented as percentages across all included studies whereas Figure 3 presents a summary of the risk of bias assessments by risk of bias domain and by trial.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
One cross‐over trial was not randomised, although the four groups of participants did receive the treatments in different orders. There was no mention of allocation concealment. We judged this trial to be at high risk of selection bias (Cohen‐Mansfield 1997). The remaining two cross‐over trials reported random allocation of treatment order, but the method of sequence generation and allocation concealment were not described and so we judged the risk of bias in these domains to be unclear (Camberg 1999; Garland 2007). Hence, none of the studies was at low risk of selection bias.
Blinding
Given the types of intervention, participants and personnel could not be blinded in any of the studies. We therefore judged all the studies to be at high risk of performance bias.
The outcome assessor was blinded in two studies and these two were therefore considered at low risk of detection bias (Camberg 1999; Garland 2007). No information regarding the blinding of the outcome assessor was reported for Cohen‐Mansfield 1997 so we judged it to have an unclear risk of detection bias.
Incomplete outcome data
Of the 60 participants included in Cohen‐Mansfield 1997, 28 participants (47%) were excluded from the analysis (seven died before any data were collected, 11 became quiet, three refused to continue, and two were physically restrained and thus excluded). We therefore considered that the trial was at high risk of attrition bias.
In Camberg 1999, five out of 54 subjects were ill and could not participate in some sessions during the study. The authors reported that the available information was retained for all analyses. We judged it and Garland 2007 to be at low risk of attrition bias.
Selective reporting
None of the studies had a protocol available. We judged the risk of bias due to selective outcome reporting in all studies to be unclear.
Other potential sources of bias
All the studies were cross‐over in design and were considered suitable for evaluating this type of intervention. However, two studies failed to report paired data and their analyses were performed on observations rather than participants, thus introducing a unit of analysis error. These studies were considered at high risk of bias (Camberg 1999; Cohen‐Mansfield 1997). We were unsure whether Garland 2007 analysed the number of observations and therefore considered the analysis at unclear risk of bias
Effects of interventions
See: Summary of findings 1 Summary of findings
Primary outcomes
1. Behavioural and psychological symptoms
Although all participants had severe dementia, the studies included people with different behavioural disturbances. In particular, Cohen‐Mansfield 1997 selected only those people with verbally disruptive behaviour. Therefore, the samples evaluated in the three studies were not comparable with each other. Moreover, behavioural and psychological symptoms were not uniformly defined across the trials and different tools were used to measure this outcome. These issues, together with the varied study designs, meant that we considered that data from the different studies could not be combined in meta‐analyses.
Despite their cross‐over design, none of the trials reported paired data. In addition, we were unable to extract the data specified in our protocol (Camberg 1999 and Cohen‐Mansfield 1997 based their analysis on number of observations and provided P values; Garland 2007 provided the mean frequencies or differences in mean behaviour counts, before and during treatments, between SPT and control, but did not report standard deviations).
We considered the overall quality of the evidence to be very low due to serious concerns regarding the risk of bias (all studies were at high risk of bias in one or more domains, including high risks of selection, performance, attrition and other bias), and very serious concern regarding imprecision (all results were based on single studies with very small sample sizes) (summary of findings Table 1).
We present below the results reported by study authors separately for each study. However, due to our quality assessment, it should be noted that we consider there to be a very high level of uncertainty about all of these results.
In Camberg 1999, staff generally used the audiotapes twice a day for each subject. Unfortunately, the audiotapes were employed when there was a pause in staff activity, instead of whenever subjects showed agitated or withdrawn behaviour.
The results were different according to the method used to assess the outcomes. Concerning direct observation, in 8000 observations by trained non‐participants, audiotapes were employed infrequently (SimPres 11.7% and placebo 10.5% of the time). Moreover, interventions were seldom fully observed, given that a predetermined schedule was used to assess behaviour, independently of whether an intervention was administered or not. The authors stated that “A Latin Squares analysis was conducted with 20,667 direct observations obtained during the study period under the "intention to treat" assumption”, which did not reveal an overall difference in either agitated or withdrawn behaviours. A further Latin Squares analysis, only of observations comprising an intervention (843 for SimPres and 780 for placebo), did not reveal any statistically significant main effects for agitation (SOAPD; PARS). In addition, analysis of frequency of happy expressions, using facial diagrams of mood (FACE) data, did not demonstrate a statistically significant difference between SimPres and usual care, but revealed that subjects had a happy expression more often during SimPres administration (11% of the time) than during placebo (2.2% of the time) (P < 0.001).
Staff observation logs documented 2547 observations of interventions for agitation by nursing staff (SimPres 592, placebo 600 and usual care 1355 observations). In contrast to the previous negative findings based on direct observation data, the authors reported that SimPres decreased agitation statistically significantly more often than placebo (P < 0.001) and usual care (P < 0.001). The same data collection method registered 1981 observations of interventions for withdrawn behaviour (SimPres 759, placebo 660, usual care 562 observations). Again, the staff reported that SimPres statistically significantly improved withdrawn behaviour twice as much as placebo (P < 0.001) and more often than usual care (P < 0.001).
Weekly staff survey data showed that there was no statistically significant difference in the amount of participants’ agitation between either SimPres and placebo or SimPres and usual care, although participants were slightly less agitated with usual care than with placebo (P = 0.017). Regarding mood evaluated with the MOSES scale, none of the interventions was statistically significantly superior. Concerning “interest” measured with the MOSES scale, there was an increase with SimPres compared to both placebo (P = 0.008) and usual care (P = 0.001) and no difference between placebo and usual care.
In Cohen‐Mansfield 1997, analysis of tape recording data revealed that VDB declined 46% during SPT (videotape recordings), 31% during music, 56% during the one‐on‐one social interaction and 16% during the “no‐intervention”, compared to before the intervention. The authors stated that post hoc tests showed that each of the interventions were significantly different from “no‐intervention”.
Standardised observational data, analysed using the SBMI for nine types of VDB, showed that compared to pre‐intervention levels, SPT (videotape recordings) was best at decreasing repeating words (58%), hallucinations (56%) and shouting (50%). The same observational data revealed that music was most effective at reducing repeating words (48%), other VDB (48%) and hallucinations (40%). The most noticeable effects of social interaction was to decrease requests for attention (94%), repeating words (70%), and shouting (66%). Various VDB declined even with “no‐intervention”, such as requests for attention (31%), repeating words (26%), hallucinations (21%), shouting (17%) and other VDB (15%).
Nurses’ assessments (informant ratings) did not find that any intervention produced a significant effect in any of the verbally agitated behaviours.
In Garland 2007, concerning physical agitation there was no statistically significant difference between simulated presence and music during the interventions, whereas simulated presence was more effective than placebo (P = 0.007) and usual care (P = 0.003). The effect of music was not different from placebo, but was better than usual care (P = 0.039). Finally, the response to placebo was not statistically significant from the effect of usual care. At 15 minutes post‐intervention, none of the treatments had a statistically significantly different effect on physical agitation. Regarding verbal agitation during the interventions, the effect of simulated presence was no different from music or placebo, but had a statistically significant action compared to usual care (P = 0.37). The response to music was indistinguishable from the effect of placebo and usual care, while placebo was more effective than usual care on verbal agitation (P = 0.03). Fifteen minutes after the interventions, simulated presence was no different from music, placebo or usual care. In contrast, the effect of music was statistically significantly different from placebo (P = 0.007), but not to usual care. Finally, the comparison between placebo and usual care did not reveal any statistically significant difference.
2. Quality of life
Quality of life was not assessed by any of the studies.
Secondary outcomes
1. Activities of daily living
Performance of activities of daily living was not assessed by any of the studies.
2. Caregiver burden
Caregiver burden was not assessed by any of the studies.
3. Dropouts
Dropouts were not reported by the studies.
Sensitivity and subgroup analysis
There were not enough data to conduct any of the planned subgroup or sensitivity analyses.
Discussion
Summary of main results
The review included three studies of differing design, involving in total 144 participants with dementia. These people had different types of behavioural disturbances. The trials investigated the effects of SPT compared to usual care, personalised music, 'placebo' or a social interaction intervention. The experimental intervention in the three studies differed in terms of the tools used to carry out the treatment and the frequency with which it was administered. The outcomes related to behavioural disturbances were multiple, measured with different scales and with different time frames, which prevented us from performing a meta‐analysis. Within each trial, the effect of SPT on behaviour, compared to usual care, was mixed and depended on the measure used. Two trials which included a personalised music intervention reported no significant differences between simulated presence and music on behavioural outcomes. Only behavioural outcomes were measured. The overall quality of the evidence was very low. We were unable to draw conclusions about the effectiveness of SPT for behavioural problems or other outcomes.
Overall completeness and applicability of evidence
The small number of trials included in this systematic review provided a very limited body of evidence on the effectiveness of SPT for dementia. There is currently a scarcity of randomised trials in this area, with only two studies using a randomised design (Camberg 1999; Garland 2007).
All the trials were conducted in nursing homes and their dates of publication ranged from 1999 to 2007. The mean age of the participants was above 70 years across all studies and women outnumbered men in all the trials. However, the number of people with vascular dementia or Lewy Body disease was limited in this review; this fact must be taken into account when interpreting the results. Most participants across the studies had moderate to severe dementia. Only one trial reported the concurrent use of psychotropic medications (Garland 2007). There is no data on comorbidity across the studies, so we do not know how this might have influenced the results.
Trial interventions showed large differences regarding the equipment used (audio or videotape), the duration of the intervention period, the selection of the contents of the audio/videotape, the frequency of administration, the washout period, and the professionals involved. Likewise, it was unclear how the optimal dosage or delivery (sessions, duration, etc.) was determined. Two studies used music and usual care as comparators (Cohen‐Mansfield 1997; Garland 2007), while one study also used a social interaction intervention (Cohen‐Mansfield 1997). Generally, usual care was not described.
There was also great variation in both the behaviour disturbances considered/evaluated and in the assessment procedures used to measure the outcomes. For example, while Camberg 1999 proposed agitation (defined as an experience of an unpleasant state of excitement) and withdrawn behaviours, composed of manifestations of mood and "interest", Garland 2007 contemplated physically aggressive and non‐aggressive agitation, and verbally aggressive and non‐aggressive agitation, but described results only for physical and verbal agitation. Conversely Cohen‐Mansfield 1997 considered only verbally disruptive behaviours defined as “verbal or vocal behaviours that are either repetitive, disruptive, or inappropriate to the circumstances in which they are manifested”.
In terms of outcome assessment, in two trials several scales were used to measure the outcome (Camberg 1999 used OAS, AVAS, SCMAI for agitation and PARS, FACE and MOSES for withdrawal behaviour; Cohen‐Mansfield 1997 used SBMI and CMAI in addition to direct observation), while Garland 2007 did not clearly report the scale used to assess physically and verbally agitated behaviours. All these factors hindered the possibility of performing a meta‐analysis, including the fact that some studies did not control for other variables such as the use of psychotropic medication.
In terms of the results, authors from each trial provided conclusions in favour of the effect of SPT on behaviour‐related outcomes. However, in some cases, the results varied depending on the outcome measured. For example, despite Camberg 1999 stating that SPT “can be effective in enhancing well‐being and decreasing problem behaviors”, SPT did not ameliorate agitation when measured with direct observations or weekly staff survey, whereas a statistically significant improvement was observed from staff observation logs. In another case, despite SPT being reported to be better than usual care, it was no more effective than personalised music (Garland 2007).
Other important outcomes of interest such as quality of life, activities of daily living, caregiver burden and dropouts were not assessed.
The most likely adverse effect we anticipated was worsening of agitation which could be detected in measures of the primary outcome. Worsening of agitation during treatment as a measure of adverse event was reported in Garland 2007. The authors of this study reported that some participants became more agitated and threw the headphones away. However, they did not provide the number of participants that experienced the outcome nor the frequency of agitation. No dropouts were reported in the other two trials.
Quality of the evidence
The quality of evidence was downgraded due to very serious concern in the domains of the risk of bias. Despite two of the studies being described as randomised, the method of sequence generation was not adequately reported and allocation concealment was not described (Camberg 1999; Garland 2007). Cohen‐Mansfield 1997 used an alternation method between interventions and was thus judged to be at high risk of selection bias. Therefore, none of the studies was rated at low risk of selection bias. In terms of blinding, none of the included studies was able to blind participants or treatment providers because of inherent difficulties in blinding non‐pharmacological interventions such as SPT. Therefore, all the studies were exposed to performance bias. Two studies reported that the outcome assessors were blinded (Camberg 1999; Garland 2007), whereas Cohen‐Mansfield 1997 did not report any statement and was thus rated unclear in terms of detection bias. During follow‐up, a significant percentage (47%) of participants were excluded from analysis in Garland 2007 and the study was judged to be at high risk of attrition bias. No study protocols were available for any of the studies to compare the outcomes declared with those evaluated; and, furthermore, details of the results of the primary outcome were not clearly reported across all the studies thus hindering the possibility of performing meta‐analyses and were therefore rated to have unclear risk of selective reporting bias.
A further downgrading by two levels of the evidence was performed for imprecision, since the studies had very few participants. In conclusion, the overall evidence for SPT to treat BPSD was rated very low in the summary of findings Table 1.
Potential biases in the review process
A protocol for this review was published (Abraha 2016; Abraha 2015). The review was conducted using comprehensive searches in several electronic databases to ensure all published trials were identified. The inclusion criteria for this review were intentionally broad to capture all studies. We did not limit the searches to a particular language and used pairs of review authors to independently examine and select studies. Authors of all the studies included were contacted to ascertain if any data were available since publication of the original studies, and to clarify unclear data. However, no responses were forthcoming.
Agreements and disagreements with other studies or reviews
One review that examined the effect of SPT in people with dementia identified seven studies (Zetteler 2008). Only three of those studies were included in our review, whereas the remaining trials were excluded, mostly due to study design. In Zetteler 2008's review, the author was able to obtain data from primary study authors, and hence to perform a meta‐analysis of four primary studies (Cheston 2007; Garland 2007; Miller 2001; Woods 1995), only one of which was included in our review (Garland 2007). The results of this meta‐analysis showed a mean effect size of 0.70, with a 95% CI of 0.38 to 1.02 (P < 0.001) with significant heterogeneity I² = 72% in favour of SPT for challenging behaviours in dementia. One study contributed most to the heterogeneity, but its removal did not affect the significance of the results (Woods 1995). However, while Zetteler 2008's review underlined the lack of statistical power in the original studies, other methodological issues, such as risk of bias, were not assessed.
Study flow diagram.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
| Simulated presence therapy compared with usual care for people with dementia | |||
| Patient or population: dementia | |||
| Outcomes | Impact | № of participants | Quality of the evidence |
|---|---|---|---|
| Overall BPSD (multidomain) – not assessed | ‐ | ‐ | ‐ |
| Agitation | One study reported that agitation assessed by two methods (direct observation and weekly staff survey) was not reduced by SPT, while a third method (staff observation logs) showed a decline in agitation. In a second study, two methods (VDB score and standardised observational data) showed that SPT reduced verbally disruptive behaviour, while there was no effect on the same outcome with a third method (nurses’ assessments). A third study with an unclear assessment method reported that SPT was effective in reducing physical agitation and verbal agitation compared to usual care. | 144 | ⊕⊝⊝⊝ |
| Depression – assessed in one study as "withdrawn behaviours" which included a lack of interest and sad mood | SPT did not improve withdrawn behaviours, interest or mood based on direct observations. Weekly staff surveys did not show any effect of SPT on mood, although the intervention was associated with increased interest. The staff, through observation logs, reported that SPT compared to usual care statistically significantly improved withdrawn behaviour. | 54 (1 trial) | ⊕⊝⊝⊝ |
| Anxiety ‐ not measured | ‐ | ‐ | ‐ |
| Quality of life ‐ not measured | ‐ | ‐ | ‐ |
| Dropouts ‐ not measured | ‐ | ‐ | ‐ |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||
| GRADE Working Group grades of evidence | |||
| 1 Downgraded by two levels because of very serious concern about the risk of bias: none of the trials was at low risk of selection bias (one study was at high risk of bias); all the trials were at high risk of performance bias; one trial was at high risk of attrition bias; all had unclear selective reporting; and two were at high risk of other biases. 2 Downgraded by two levels because of serious concern about imprecision: none of the studies reached the optimal information size. 3 Downgraded by two levels because of serious concern about the risk of bias; the trial was at unclear risk of selection bias; high risk of performance bias; unclear risk of selective reporting bias; and high risk of other bias. 4 Downgraded by two levels because of very serious concern about imprecision: very small sample size. | |||
