Non-pharmacological interventions for Lewy body dementia: a systematic review

Eligibility criteria

The review focused on primary research that assessed a non-pharmacological intervention for either patients with Lewy body dementia or carers of such patients. The review considered studies evaluating non-pharmacological interventions in the broader categories of dementia or Parkinson's disease if the results of a Lewy body dementia subgroup were available. Studies that were confounded by a concurrent pharmaceutical intervention (i.e. medication initiated at the same time as a non-pharmacological intervention or given to one group in a study but not the other) were excluded. There were no other restrictions on study design and no requirements for a comparator group. There were also no restrictions on language, time period, or clinical context in which the study was conducted.

Search strategy

The search identified studies through bibliographic databases, trial registers, and the grey literature. Bibliographic databases and trial registers included the following: Medline (1946–present); PreMedline, PubMed; EMBASE (1974–present), Scopus, Web of Science (1900–current); PsychInfo (1806–present); CINAHL (1981–present); Cochrane libraries: Cochrane database of systematic reviews (2005–October 2016), Cochrane central register of controlled trials (August 2016), Cochrane Methodology register (3rd–Quarter 2012); other EBM databases: ACP journal club (1991–September 2016), Database of Abstracts of Reviews of Effects (1st-quarter 2015), Health technology assessment (3rd-quarter 2016), and NHS economic evaluation (1st-quarter 2015); Ageline (1978–present); ALOIS; AMED (Allied and Complementary Medicine; 1985–present); PEDro (Physiotherapy Evidence Database; 1929–present); Social work Abstracts (1968–present); and the National Association of Social Workers (NASW) clinical register (14th edition). The grey literature was searched using such resources as SIGLE (System for Information on Grey Literature in Europe), NTIS (National Technical Information Service) database, and PsychEXTRA (1908–present).

The search strategy used only population and intervention terms to maximise the likelihood of identifying relevant studies (comparator and outcome terms were not used). The population was people with Lewy body dementia or their carers. This was identified using the search terms: [(Lewy OR Park*) and Dementia]. Interventions were any non-pharmacological treatment and identified using a wide range of terms: (activit*, acupuncture, alternative, animal, aromatherapy, art therapy, assisted, balance, behav*, bicycle, calisthenics, carer intervention, caregiver intervention, CBT, Chi gong, cognit*, cognitive behavioral therapy, cognitive behavioural therapy, counsel*, creative arts, dance, dancing, diet, direct current stimulation, drama, ECT, educat*, electroconvulsive therapy, enhanc*, environmental intervention, environmental modification, exercise, flexibility, humor therapy, humour therapy, hydrotherapy, intervention*, leisure, light therapy, management, martial arts, massage, meditation, Montessori, multisensory, music, non-pharm*, nonpharm*, nutrition, occupational therapy, pet therapy, physical activity, physical therapy, physiotherapy, pilates, psychoeducation, psychol*, psychosocial, psychotherapy, Qi gong, reality orientation, recreation*, reminiscence, resistance training, run*, sensory, simulated presence, stimulation, Snoezelen, support*, support group*, swim*, tai chi, therap*, therapeutic activity, TMS, training, training carers, training caregivers, transcranial magnetic stimulation, treatment*, validation, weight training, yoga). Searches were conducted on 30 October 2016.

In addition to bibliographic database searches, the reference lists of papers included in the review and previous systematic reviews on both Lewy body dementia and non-pharmacological interventions were checked for relevant papers. Advice was also sought from experts in the field.

Study selection

Two reviewers (MHC and LQ) independently assessed search results for inclusion by title and abstract. All articles deemed relevant by either reviewer were obtained in full. Both reviewers then independently evaluated full-text articles for inclusion. Any disagreements were resolved through discussion or, if necessary, with a third reviewer (JTO).

Data extraction

Two reviewers independently extracted relevant data from publications using a standardised form. This included participant details (e.g. demographics, number, recruitment, clinical context, dementia severity), intervention type, study design, measures, and results. Qualitative data were also collated.

The primary outcomes were measures of cognition, function, neuropsychiatric symptoms, and motor symptoms. The secondary outcomes were measures of any other clinically relevant outcomes, such as quality of life, carer burden, financial costs, other symptoms (sleep or autonomic disturbances), and objective endpoints (e.g. falls, hospitalisation, institutionalisation, mortality). Secondary outcomes also included the perceived acceptability of treatments, reported side effects, and dropout rates (a measure of treatment acceptability).

Quality assessment

Two reviewers independently assessed study quality and risk of bias using standardised tools. These included the Effective Public Health Practice Project Quality Assessment Tool for Quantitative Studies (Effective Public Health Practice Project, 2007; Armijo-Olivo et al. 2012) and the NICE Methodology Checklist: Qualitative Studies (NICE, 2016). Any disagreements were resolved through discussion.

Data synthesis

Although a quantitative synthesis of findings using meta-analytic techniques was originally intended, this was not possible due to the small sample sizes and poor quality of the included studies. As a result, only a descriptive synthesis was provided.

Search results

The search identified 73 093 publications, of which 30 419 were unique and 42 674 were duplicates. Of these, 76 were considered by one or both reviewers to be potentially relevant and obtained in full. In turn, 21 studies were found eligible for inclusion (a flow diagram in PRISMA format is shown in Fig. 1). The included studies consisted of 12 case studies (Kung & O'Connor, 2002; Loher et al. 2002; Graff et al. 2006; Cheston et al. 2007; Huh et al. 2008; Freund et al. 2009; Ciro et al. 2013; Gil-Ruiz et al. 2013; Tabak et al. 2013; Dawley, 2015; Hsu et al. 2015; Ricciardi et al. 2015), seven case series (Rasmussen et al. 2003; Rochester et al. 2009; Takahashi et al. 2009; Ota et al. 2012; Elder et al. 2016; Yamaguchi et al. 2016) – two of which were reported in the same paper (Takahashi et al. 2009) – and two randomised controlled trials (Logemann et al. 2008; Telenius et al. 2015). Both randomised trials focused on the broader conditions of dementia and/or Parkinson's disease but had separate results for a Lewy body dementia subgroup available. One case report was reported in two separate papers (Freund et al. 2009; Barnikol et al. 2010); our descriptive synthesis focused on the more detailed of these papers (Freund et al. 2009). A further case report on electroconvulsive therapy was excluded (Fujiwara et al. 2004) because of a concurrent pharmacological intervention. Twelve studies focused on dementia with Lewy bodies (Kung & O'Connor, 2002; Rasmussen et al. 2003; Cheston et al. 2007; Huh et al. 2008; Takahashi et al. 2009; Ota et al. 2012; Ciro et al. 2013; Gil-Ruiz et al. 2013; Hsu et al. 2015; Telenius et al. 2015; Yamaguchi et al. 2016), eight focused on Parkinson's disease dementia (Loher et al. 2002; Graff et al. 2006; Logemann et al. 2008; Freund et al. 2009; Rochester et al. 2009; Tabak et al. 2013; Dawley, 2015; Ricciardi et al. 2015), and one included patients from both groups (Elder et al. 2016). A summary of studies is shown in Table 1.

Quality assessment

Of the 21 studies, 19 were rated as poor quality due to their small sample size, uncertainties about recruitment, and the lack of a control group. One randomised controlled trial, which focused on exercise in nursing homes for patients with dementia (Telenius et al. 2015), was evaluated as poor quality with respect to Lewy body dementia specifically (it only included four participants with the condition and these participants did not complete all tasks) (Inskip et al. 2016). The other randomised controlled trial (Logemann et al. 2008), which investigated dietary fluid and postural interventions for dysphagia, was evaluated as moderate quality. This study, however, was limited by its lack of blinding and the fact that it relied on measures of immediate efficacy in a research setting, without any longer term follow-up or measures of real-world effectiveness. It was also limited by the absence of a control group that did not receive a treatment, making it difficult to establish the absolute effectiveness of the interventions.

Participants

The total number of participants across the included studies was 195. This comprised 44 patients with dementia with Lewy bodies and 151 patients with Parkinson's disease dementia (of whom, 132 were from one study; Logemann et al. 2008). Demographic information was not consistently reported. It was not provided for the subgroup of 132 patients with Parkinson's disease dementia in the largest study (Logemann et al. 2008) and for two case series (Ota et al. 2012; Yamaguchi et al. 2016) – altogether 140 participants. Of the data available (n = 55), the mean age of participants was 70.6 years (s.d. = 8.2) and included 27 males and 28 females. Measures of patients’ cognition and dementia severity were lacking from the majority of reports. Of the 21 studies, nine recruited participants from nursing homes or hospitals, five recruited participants from the community, and seven did not report participants’ residential status. One of the studies that recruited participants from nursing homes examined the effect of an intervention on both a patient with dementia with Lewy bodies and her carers (Huh et al. 2008).

Interventions

Interventions included carer education (Huh et al. 2008), psychological interventions (for visual hallucinations) (Ota et al. 2012), physical exercise (Tabak et al. 2013; Telenius et al. 2015), gait cueing (Rochester et al. 2009), environmental modification (Gil-Ruiz et al. 2013) (for mirrored self-misidentification, a delusion that usually occurs in the context of dementia; Connors & Coltheart, 2011; Connors et al. 2015), music (Hsu et al. 2015), simulated presence (Cheston et al. 2007), occupational therapy (Graff et al. 2006; Ciro et al. 2013), physical therapy (Dawley, 2015), dietary fluid and postural interventions to prevent aspiration (Logemann et al. 2008), electroconvulsive therapy (Kung & O'Connor, 2002; Rasmussen et al. 2003; Takahashi et al. 2009; Yamaguchi et al. 2016), transcranial magnetic stimulation (Takahashi et al. 2009), transcranial direct current stimulation (Elder et al. 2016), and deep brain stimulation (Loher et al. 2002; Freund et al. 2009; Ricciardi et al. 2015). The most studied interventions were electroconvulsive therapy (four studies) and deep brain stimulation (three studies).

Effectiveness of interventions

Given the heterogeneity of interventions and the poor quality of the research evidence, no quantitative synthesis was possible. A descriptive summary of individual studies is provided in Table 1.

All studies reported some effectiveness of their respective intervention. The strongest evidence came from the randomised control trial that assessed interventions to prevent fluid aspiration in participants with dysphagia (Logemann et al. 2008). This trial compared honey-thickened fluids, nectar-thickened fluids, and a chin-down posture in 132 patients with Parkinson's disease dementia (see Table 1). It found that the use of honey-thickened fluids was superior to the other two methods in preventing fluid aspiration as measured by videofluorographic swallow studies.

Four small uncontrolled studies (Kung & O'Connor, 2002; Rasmussen et al. 2003; Takahashi et al. 2009; Yamaguchi et al. 2016) – involving 22 participants in total – found that electroconvulsive therapy had some effectiveness in treating depression in dementia with Lewy bodies. In one of these studies, two patients exhibited confusion immediately after electroconvulsive therapy (Rasmussen et al. 2003); in another study, an unspecified number displayed signs of autonomic dysfunction, though without any lasting effects (Takahashi et al. 2009). No other adverse events were reported. Another three case studies of deep brain stimulation (Loher et al. 2002; Freund et al. 2009; Ricciardi et al. 2015) identified benefits in either cognition or motor symptoms depending on the location of stimulation in patients with Parkinson's disease dementia. All three studies used different brain locations for stimulation. None of these studies reported adverse effects.

Other individual studies variously reported benefits of carer education for reducing agitation (Huh et al. 2008); psychological interventions for reducing distress around visual hallucinations (Ota et al. 2012); physical exercise to improve gait and function (Tabak et al. 2013; Telenius et al. 2015); gait cueing to improve gait speed (Rochester et al. 2009); environmental modification to reduce the distress associated with mirrored self-misidentification delusion (Gil-Ruiz et al. 2013); music therapy to reduce agitation (Hsu et al. 2015); simulated presence to reduce distress (Cheston et al. 2007); occupational therapy interventions to improve functional ability (Graff et al. 2006; Ciro et al. 2013); physical therapy to improve gait (Dawley, 2015); dietary fluid and postural interventions to prevent aspiration (Logemann et al. 2008); transcranial magnetic stimulation to reduce depression (Takahashi et al. 2009); and transcranial direct current stimulation to improve attention (Elder et al. 2016) (see Table 1). For all these studies, however, limitations in their design, including small sample sizes, lack of blinding, and lack of adequate controls, mean that it is not possible to rule out confounding, selection bias, experimenter expectancy, publication bias, and other causes for the reported effects.