Patient-Level Interventions for Agitation/Aggression in Nursing Home and Assisted Living Facilities

Michelle Brasure; Eric Jutkowitz; Erika Fuchs; Victoria A Nelson; Rosalie A Kane; Tetyana Shippee; Howard A Fink; Tonye Sylvanus; Jeannine Ouellette; Mary Butler; Robert L Kane

Appendix CPatient-Level Interventions for Agitation/Aggression in Nursing Home and Assisted Living Facilities

Table C1

Patient-level interventions for agitation/aggression in nursing home and assisted living facilities: risk of bias assessments.

Patient-Level Interventions for Agitation/Aggression in Nursing Home and Assisted Living Facilities: Description of Trials Rated High Risk of Bias

Music

Eight studies of music intervention were rated as having high risk of bias.⁸^,¹⁷^,²⁰^,⁴³^,⁴⁶^,⁵⁴^-⁵⁶ Both the Sung studies were participatory with the first involving movement, and the second involving percussion instruments;⁵⁴^,⁵⁵ Narme et al., was a group music therapy intervention.⁴³ The other five studies used individual interventions;⁸^,¹⁷^,²⁰^,⁴⁶^,⁵⁶ Clark et al. used pre-recorded soothing music for residents with a history of aggression during bathing; Hicks-Moore et al. used pre-recorded music participants' indicated they enjoyed paired with a hand massage; Gerdner et al. compared listening to recordings of preferred music rather than recordings of classical music; Svansdottir et al. used a music therapist to engage patients in singing and instrument-playing compared to an undefined control; and Ridder et al. tested the effects of individual in-person music therapy compared with usual care (which often included group music therapy).⁴⁶ These studies are briefly summarized below.

In a cross-over design with 18 subjects, Clark at el. examined the effects of prerecorded music on aggressive behavior among people with severe Alzheimer's type dementia during bath time in a 2-week period, compared with usual care.⁸ Significant decreases were found in hitting behaviors during the intervention, and “discussions with caregivers” was associated with less agitation during the intervention.

Hicks-Moore et al., in an unblinded, repeated-measures design, reported randomizing 41 residents with mild to moderate dementia to music, hand massage, both combined, or control.²⁰ The actual number randomized was unclear, as was attrition. Participants in the music conditions received 10 minutes of pre-recorded music they indicated was their favorite song, artist, or type. There were no significant differences between groups for aggression at posttreatment.

Gerdner, et al., in a cross-over design, 39 residents participated in a study comparing 30 minute periods listening to recordings of relaxing classical music versus recordings of preferred music during their lives, as determined by family members completing a preferred music questionnaire on their behalf.¹⁷ Dose was 30 minutes 2 days a week. The outcomes were measured by the Temporal Pattern in Assessment of Agitation (TPAA) scale, which was modified from the CMAI. The study compared the immediate and 30-minute residual effects of the individualized music. Positive findings are reported, but the raters of outcomes were the ones who applied the intervention.

Sung, et al. studied a 1-month study of 30-minute, twice weekly sessions of group music therapy with movement compared with usual care in a single large Taiwanese nursing home.⁵⁴ The authors reported significant decrease in episodes of agitation by week 2 and week 4 using CMAI.

Sung et al. randomized 60 residents from a Taiwanese residential care facility to the intervention (active participation in music therapy with percussion instruments and exercise for 30 minutes twice-weekly for a month versus usual care).⁵⁵ Authors reported no differences in agitation but significantly less anxiety on the RAID measure in the music group compared with the control group.

Narme, et al. randomized 48 residents with dementia in a single nursing home in France to music therapy or a cooking group;⁴³ 37 remained in the study for analysis. Groups lasted for an hour, and were conducted twice weekly for 4 weeks. They found no differences in reduction of agitation between the new groups measured by CMAI and by NPI.

Svansdottir et al. randomized 38 residents with moderate to severe Alzheimer's disease in four locations in Iceland to small-group music therapy or an undefined control. Residents in the intervention were either engaged in singing and instrument-playing with a certified music therapist or sat listening for 30 minutes thrice weekly for 6 weeks. The authors reported significant decreases in activity disturbances during the intervention, but no lasting effects.

Ridder et al. conducted a cross-over trial in 14 nursing homes in Denmark and Norway.⁴⁶ Forty-two paired participants were randomized to 6 weeks of individualized music therapy or 6 weeks of usual care, which could include group music therapy. In this nonblinded study, the experimental group experienced a significant reduction in agitation while the control group was reported to have had a significant increase in psychoactive medication prescriptions.

Massage

Hicks-Moore et al., in an unblinded, repeated-measures design, reported randomizing 41 residents with mild to moderate dementia to hand massage, music, both combined, or control.²⁰ However, the actual number randomized was unclear, as was attrition. Based on a previously-developed protocol, participants in the hand massage group slow, light pressure applied to each hand for five minutes. There were no significant differences between groups for aggression at posttreatment.

Aromatherapy

Smallwood et al. randomized 21 district general hospital ward patients into three groups: aromatherapy and massage, conversation and aromatherapy, and massage only (seven per group).⁵² The intervention is not well explained, but it appears that the aromatic oil was used for massage in the combined group, conversation occurred during aromatherapy for the second group, and ordinary oil was used for massage in the last group. Each individual received treatment twice weekly, after which the patients' behavior was recorded. Treatment time of day was rotated in each period so that over the course of the study each person received treatment twice in each period of the day. The study used a single-blind design. Frequency of behaviors was based on daily recordings. Findings showed no overall difference in frequency of behavior across groups. Aromatherapy and massage showed a reduction in the frequency of excessive motor behavior (one of the domain on the scale) of all three conditions which reached statistical significance between 3 p.m. to 4 p.m. (P<0.05).

Acupoint

Mariko et al., randomized 23 residents with moderate to severe dementia to either acupoint touch twice daily for four weeks, a or control.³⁹ Randomization, blinding, and analysis methods were not reported. Antipsychotics were prescribed as needed, but unclear for which groups. General behavior symptoms decreased significantly from baseline in the intervention condition, but not in control. Comparisons were not reported between groups.

Tailored Interventions

Three studies involving individual assessments and tailored activity interventions to reduce agitation were identified but rated as having a high risk of bias. One is an earlier study of the TREAS model;¹⁰ a later study of TREAS was rated as having lesser risk of bias and is included in our analysis.⁹ The second selected activities tailored towards patients' skill level, personality, and interests.²⁷ The third is a study which tailored an intervention to optimize a mix of simulation and withdrawal.³² The three studies are briefly described below.

Cohen-Mansfield et al. tested the efficacy of an algorithm for providing individualized nonpharmacological approaches to reduce agitation tailored to individual profiles of each resident's unmet needs, physical, cognitive, and sensory abilities; and with interventions based on residents' lifelong habits and roles as well as abilities: TREA (Treatment Routes for Exploring Agitation).¹⁰ Interventions were applied for 10 days during the 4 hours of the individual's greatest agitation. The study was conducted in 12 Maryland nursing homes, 6 used as experimental and 6 used as controls. The implementation of personalized, nonpharmacological interventions resulted in statistically significant decreases in overall agitation in the intervention group relative to the control group from baseline to treatment and implementation of individualized interventions for agitation resulted in statistically significant increases in pleasure and interest.

Kolanowski et al. tested the efficacy of intervention activities based on validated cognitive and personality assessments versus control activities (domestic activities such as sewing cards, hanging laundry) to a random order by having 10 residents with dementia in one nursing home serve as their own control.²⁷ Activities were performed at least 15 minutes a day for one week. The authors reported fewer dementia behaviors observed on intervention days compared to control days, although this was not significant.

Lawton et al. randomized residents from two Dementia Special Care Units in the same nursing home to the condition of receiving a package of care according to individually assessed needs for stimulation or release from stimulation (retreat).³² The study was conducted over 2 years, with considerable difficulty in implementation because of noncooperation of care teams and interference of prescribed the stimulation-retreat cycle with staff duties and resident schedules. Over time most functions worsened for both groups, agitated behavior did not decline more in the experimental unit, and there was marginal improvement in external engagement and lesser declines in positive affect and greater increases in negative affect in the experimental group.

Family Involvement in Care

Jablonski et al. tested family involvement in care using contracts to identify the type, frequency, and duration of involvement and activity that the family agreed to have.²⁶ The intervention is the Family Involvement in Care (FIC) protocol, whereby a primary family member is oriented to the facility, educated on potential involvement in resident care, and contracts to participate in a specified number of care activities in nine possible areas of care) for a specified amount of time. The dosage is calculated across all types and amount of activities. The experimental group exhibited less global deterioration but inappropriate behavior remained the same.

Creative Activity Program (TimeSlips)

Houser et al. tested a creative story telling intervention called TimeSlips.²¹ This small pilot study evaluated the creative story-telling activity known as TimeSlips (wherein residents react to a picture with story ideas that are recorded and then read back to participants as their collective story) for its effect on behavioral symptoms and mood. The intervention group of 10 residents received two 1-hour TimeSlips sessions for 6 weeks and the comparison group of 10 residents received standard activity programming for 6 weeks. In this pilot study no statistically significant differences in mood or behavior were found.

Validation Versus Sensorial Reminiscence Versus No Treatment

Deponte et al. compared validation therapy to sensorial reminiscence to no control and measured outcomes with the NPI.¹²

Simulated Presence

Garland et al. tested simulated family presence (15-minute audiotapes by a family member about a positive experience from the past), music preferred by the resident in earlier life, and a placebo condition of reading from a horticultural text, to usual care.¹⁶ The tapes were applied once a day for 3 days a week for 3 weeks. Family presence and preferred music both led to reduced counts of physically agitated behavior, and simulated presence (but not music) resulted in significantly reduced counts of verbally agitated behaviors. The placebo tape also was associated with benefits over usual care.

Hatakeyama et al. tested an intervention consisting of modified television watching by screening a person's home-made DVD with favorite pictures and greetings of family members.¹⁸ Patients in a large Japanese long-term care setting who had a dementia diagnosis participated and were assigned to a homemade or comparable length commercial DVD for 2 hours each afternoon, for 4 weeks. Positive results in agitation are reported on the NPI.

Multisensory Stimulation

Staal et al. compared multisensory behavior therapy with a structured activity session.⁵³ The study took place on a geriatric psychiatric unit using a single-blinded, between-group study design. Twenty-four participants were randomized to MSBT or structured activity. Outcomes included the Pittsburgh Agitation Scale and the Scale for the Assessment of Negative Symptoms in Alzheimer's disease. Combination treatment of MSBT and standard psychiatric care reduced agitation and apathy more than standard psychiatric inpatient care alone (P = 0.05). Multiple regression analysis predicted that within the multisensory group, apathy and agitation were reduced (R2 = 0.42; p = 0.03).

Milev et al. used multisensory stimulation (MSS) study (using a Snoezelen room), in this case a dimly lit room that included many objects pertaining to the five senses: fiber-optic cables, aroma therapy, different music/sounds, water columns of different colors, textured balls to touch, and screen projectors, among others.⁴² Subjects were assigned to one of three groups. The control group received no experimental treatment for the entire duration of the study and had only care as usual. The first experimental group had one Snoezelen session per week, and the other experimental group had three Snoezelen sessions per week for 12 weeks. Each session lasted for 30 minutes on a 1:1 basis with a qualified Snoezelen facilitator. At the end of the 12 weeks, all participants received no Snoezelen treatment for another 12 weeks. The 21 participants were randomly assigned to one of three groups. Outcomes included DOS mean scores. Patients who received one and three Snoezelen treatments per week had a consistently lower DOS mean score (i.e., they improved), without much fluctuation when compared with the control group. The effect was sustained even 12 weeks after the cessation of intervention.

Bathing

Sloane et al. randomized residents with dementia and a history of agitation during bathing to person-centered showering, a towel bath (i.e., a person-centered, in-bed, bag-bath with no-rinse soap), or usual care bathing.⁵¹ The study was done in nine Oregon and six North Carolina facilities using a cross-over design between the two experimental conditions with randomization at the facility level. The Care Recipient Behavior Assessment (CAREBA), a modification of the CMAI, was used to rate behaviors for the videotaped bathing experience. All measures of agitation and aggression declined significantly in both treatment groups but not in the control group, with aggressive incidents declining 53 percent in the person-centered shower group (P<.001) and 60 percent in the towel-bath group (P <.001). Discomfort scores also declined significantly in both intervention groups (P <.001) but not in the control group. The two interventions did not differ in agitation/aggression reduction.

Multisensory Stimulation Versus Reminiscence

Baillon, et al. used Snoezelen versus reminiscence sessions as an attention control.² Each subject was allocated one of three research staff with whom they had all their intervention sessions. This staff member spent time with the resident prior to commencing the interventions. Sessions lasted up to 40 minutes every day for 2 weeks. The study was done at the Bennion Centre, Glenfield General Hospital, at Foxton Grange, which is a charity-run nursing home for older people, and at the Evington Centre, Leicester General Hospital. Subjects were randomized to one of two groups using a sealed envelope technique. Outcomes included the ABMI with reference to 3-minute samples before, immediately after, 15 minutes after, and 30 minutes after each therapy session. No statistically significant differences were seen between Snoezelen and Reminiscence sessions in terms of the change in level of agitation from pre-session to immediately post-session (CI -4.3 to 2.0) or from pre-session to 15 minutes post-session (CI -2.0 to 3.4).

Exercise

Landi et al. studied and exercise program in nursing homes in managing dementia residents' behaviors and use of antipsychotic drugs.³¹

Therapeutic Touch

Woods et al. studied therapeutic touch on behavior of nursing home residents with dementia.⁶³

Table C2

Patient-level interventions for agitation/aggression in nursing home and assisted living facilities: strength of evidence assessments.

Publication Details

Copyright

Publisher

Agency for Healthcare Research and Quality (US), Rockville (MD)

NLM Citation

Brasure M, Jutkowitz E, Fuchs E, et al. Nonpharmacologic Interventions for Agitation and Aggression in Dementia [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2016 Mar. (Comparative Effectiveness Reviews, No. 177.) Appendix C, Patient-Level Interventions for Agitation/Aggression in Nursing Home and Assisted Living Facilities.

Table C1Patient-level interventions for agitation/aggression in nursing home and assisted living facilities: risk of bias assessments

Study	Risk of Bias Assessment
Ancoli- Israel, 2003¹	Moderate - Patient blinding unclear; staff/outcome assessors not blinded, staff reported preconceptions of what each treatment group would do for the patients; analysis methods do not mention attrition, but period is short so possibly little.
Baillon, 2004²	High - High performance, detection, and attrition bias.
Baker, 2003³	Moderate - Fidelity issues, also different organizational contexts across countries, high attrition.
Ballard, 2002⁴	Moderate - Randomization at facility level but study does not account for facility differences.
Beck, 2002⁵	Moderate - Attrition not equal in groups.
Burns, 2011	Moderate - High attrition at posttreatment; not corrected for multiple comparisons.
Burns, 2009⁶	Low
Camberg, 1999⁷	Moderate - Intervention not implemented as directed; instruments not validated for agitation; staff reporting may introduce bias.
Clark, 1998⁸	High - Blinding unclear; only valid data presented in line graphs; other data combines from crossover groups; attrition and missing data unclear.
Cohen-Mansfield, 2012⁹	Moderate - Selection bias; unclear performance bias; potential detection bias.
Cohen-Mansfield, 2007¹⁰	High - Partial randomization; baseline characteristics not similar regarding age; no mention of blinding of participants, interventionists, manuals, power analysis, attrition, or handling of missing data; outcome assessors not blinded; co-interventions not similar.
Cooke, 2010¹¹	Low
Deponte, 2007¹²	High - Selection and randomization unclear; high performance bias; blinding and fidelity unclear; incomplete data not handled appropriately; underpowered.
Dowling, 2007¹³	Moderate - Performance and detection bias unclear; high attrition.
Fu, 2013¹⁴	Moderate - Mid intervention they had 5 dropouts that withdrew consent because they wanted to be sure they were in experimental group--they were dropped so ITT model not completely used.
Fuji, 2008¹⁵	Moderate - Performance bias may be an issue; they did the aroma therapy 3 times a day, an hour after meals; no placebo control.
Garland, 2007¹⁶	High - Unclear randomization method and baseline characteristics; no mention of fidelity checks, manuals, outcome assessors, power and attrition; high risk of reporting bias; crossover study, unclear if patients were observed for all outcomes or only those which the patient primarily displayed; unsure if 2-day washout is appropriately long enough, assessors not completely blinded (seemed to guess which treatment the participant had); many excluded participants seemingly after randomization; not ITT analysis.
Gerdner, 2000¹⁷	High - Unclear randomization method, no mention of blinding of participants, interventionists, fidelity checks, manuals and outcome assessors; no mention of power analysis or handling of missing data; crossover study, unblinded outcome assessor (RA who did assessment was also there while the music was playing); possibly unbalanced groups at baseline (no demographic table, but mention 2 of 16 demographic variables significantly different); low attrition; not ITT analysis.
Hatakeyama, 2010¹⁸	High - Small sample size; selective recruitment unclear randomization, blinding, attrition, fidelity.
Hawranik, 2008¹⁹	Moderate - Selection bias, small sample size, diffusion, definition of intervention, fidelity unclear.
Hicks-Moore, 2008²⁰	High - Randomization and allocation methods unclear, unblinded, power calculation not reported, attrition unclear.
Houser, 2014²¹	High - Unclear randomization method; small study sample (no power analysis); no mention of blinding of participants, interventionists, fidelity checks, manuals and outcome assessors; no mention of handling of missing data.
Hozumi, 1996²²	Moderate - Participants seemingly blinded, unclear about outcomes assessors; attrition unclear; missing data unclear.
Hutson, 2014²³	Moderate - Unclear if participants in control group received treatment, unblinded, underpowered at randomization.
Ito, 2007²⁴	Moderate - Participants and staff not blinded; high attrition and MNAR obvious from group comparisons; not ITT analysis; high possibility for bias for how data is presented for primary and secondary analysis purposes.
Janata, 2012²⁵	High - Unblinded, power calculation not reported, most results reported via graphs.
Jablonski, 2005²⁶	High - Selection bias; performance bias actual implementation of intervention, fidelity checks, hard to know what exactly was done for the intervention.
Kolanowski, 2001²⁷	High – Randomization methods not described, patients served as their own controls, small sample size, power calculation not reported.
Kolanowski, 2005²⁸	Moderate - Crossover study; outcome assessors blinded; only selected certain behaviors by patient; did not assess all behaviors; fidelity checks appropriate; study design and analysis very confusing since A, B, and C treatments are individualized; multiple comparisons correction unclear but unlikely and many comparisons were made, not ITT analysis.
Kolanowski, 2011²⁹	Moderate - Baseline differences among groups.
Kovach, 2004³⁰	Moderate - Performance bias; nurses sometimes didn't implement if they were too busy; had to change schedules several times, groups unequal at baseline; high attrition.
Landi, 2004³¹	High - Authors call this a case control study but mention randomization; blinding unclear; little description of intervention, no mention of attrition or missingness; bar graphs only for outcomes.
Lawton, 1998³²	High - About 50% attrition rate, missing data not appropriately handled, outcome assessors not blinded; fidelity and power unclear.
Lichenberg, 2005³³	Moderate -Assessor not blinded—an outside geriatric neuropsychologist; attrition not reported.
Lin, 2007³⁴	Moderate - Non-random sampling, issues with design (assumes no changes in condition), staff were not blinded to outcomes.
Lin, 2009³⁵	Low
Lin, 2011³⁶	Moderate - Some issues with selection bias and contamination (in the same facility); blinding unclear; low attrition; likely not ITT analysis, but unclear; did not appear to correct for multiple comparisons.
Low, 2013³⁷	Moderate - Theory-base unclear, unclear if manualized, outcome assessor blinding compromised.
Lyketsos, 1999³⁸	Moderate - High attrition, not clear about how these patients were selected (and small sample size), whether sample was appropriate (did not report concerns with sleep/wake cycles), concerned about diffusion, not clear if staff were trained differently, etc.
Mariko, 2015³⁹	High - Randomization and allocation methods not described, standardization of intervention unclear, missing data handling unclear.
Maseda, 2014	Moderate - Unblinded outcome assessor, power calculation not reported, all results reported via graphs.
McCallion, 1999a⁴⁰	Moderate - Unclear regarding method of randomization and study may be under powered, unbalanced on two baseline measures one being length of stay which may mean there are unobserved disease severity variables impacting results.
McGilton, 2003⁴¹	High – Issues with selection bias, unclear intervention intensity and fidelity to treatment, small sample size.
Milev, 2008⁴²	High - No ITT analysis; mostly unblinded; low attrition, but small population; unbalanced groups at baseline; selection bias, contamination, power issues, inadequate randomization. -Incomplete data not handled appropriately; lack of blinding of outcome assessors.
Moyle, 2014	Moderate - Some detection bias (underpowered, multiple comparisons not corrected for)
Narme, 2014⁴³	High - Selection bias (only native French speakers, those without musical expertise, etc.); diffusion issues in same NH; attrition issues, small sample size, no usual care control group; very high attrition; assessors blinded; blinding of participants unclear; not ITT analysis.
Raglio, 2010⁴⁴	Moderate - High attrition issues, differences across experimental and control groups, inadequate controls in statistical models.
Remington, 2002⁴⁵	Moderate - Issues with detection bias, questions about whether this was the right sample (residents all had low scores for agitation).
Ridder, 2013⁴⁶	High – Unblinded; some baseline differences; paired RCT; say ITT analysis, but then say they exclude missing data from main analysis.
Robichaud, 1994⁴⁷	Moderate - Issues with selection bias, sample size, diffusion across institutions, concerns about fidelity of the program; some baseline differences between groups; second assessment unblended; participant blinding unclear; ITT analysis; low attrition.
Rodriguez-Mansilla, 2013⁴⁸	Moderate - Self designed instruments not validated.
Rolland, 2007⁴⁹	Moderate - Only assessor blinded; low attrition; not ITT analysis for outcomes we are interested in here.
Sakamoto, 2013⁵⁰	Moderate - Not enough information about selection of patients, short followup, power issues; participant blinding unclear; assessors blinded; corrected for multiple comparisons; attrition unclear.
Sloane, 2004⁵¹	High - Group-randomized trial; participant blinding unclear; assessors blinded; analyses combined treatment groups (person-centered vs. towel were separate time periods in two groups, seemingly combined for analysis); baseline differences for important characteristics; corrected for multiple comparisons; attrition unclear.
Smallwood, 2001⁵²	High - Study does not have enough power to detect a difference; the allocation of subjects is poorly described; differences between controls and intervention group; participant blinding unclear; outcome assessors and aromatherapist blinded; attrition unclear; analysis methods not described.
Staal, 2007⁵³	High - Selection bias, low sample size, differences between exp. and control groups; participant blinding unclear; nurse assessors unblinded; reported significant group differences at baseline; attrition unclear.
Sung, 2006⁵⁴	High - Low attrition, but not ITT analysis; researchers blinded, outcome assessors not blinded.
Sung, 2012⁵⁵	High - No blinding; single facility; tool for measuring anxiety has poor validity; low attrition, but not ITT analysis; not blinded.
Svansdottir, 2006⁵⁶	High - Authors refer to it as a case-control study, but it appears to be RCT; little information about analysis, but not ITT analysis according to tables; Attrition under 20%; outcome assessors blinded; blinding of intervention staff and patients unclear.
Telenius, 2015⁵⁷	Moderate - Multiple comparisons not corrected for; some details of standardization and fidelity to intervention unclear.
Van de Winckel, 2004⁵⁸	High - Practitioner not blinded; behavior assessors blinded to treatment; cognition assessor not blinded (same person who delivered intervention); low attrition.
Van der Ploegg, 2013⁵⁹	Moderate - Control intervention same number of one-on-one therapy time; unclear if outcomes assessors blinded; underpowered; odd selection of instruments for agitation trial.
Van Haitsma, 2015⁶⁰	Moderate – Randomization methods unclear, blinding methods unclear, unclear if outcome measures validated, power calculation not reported.
Vink, 2013⁶¹	Moderate - Nurses who took patients to activities were those who completed outcomes instruments; not ITT analysis; attrition okay, but excluded a lot of people from analysis.
Woods, 2005⁶²	Moderate - Blinded patients and assessors, not research assistants who performed intervention; questionable assignment of research assistants to intervention and placebo groups.
Woods, 2009⁶³	High – Blinded; dropped one participant from the study and analysis due to problem behaviors; did not correct for multiple comparisons, yet many time points shown; did not seem to present results for each measure collected.
Yang, 2015⁶⁴	Moderate - Randomization and allocation methods not described, investigators not blinded, moderate level of attrition.

Table C2Patient-level interventions for agitation/aggression in nursing home and assisted living facilities: strength of evidence assessments

Comparison	Outcome (Instrument) # Trials (n)	Summary Statistics	Study Limitations	Directness	Precision	Consistency	Reporting Bias	Strength of Evidence
Music vs. No treatment/Attention Control (for sustained reduction in agitation/aggression)	Patient Agitation/Aggression k=4; n=233	Standardized Mean Difference³⁶^,⁵⁰ -0.18 95% CI:-2.41 to 2.05 NPI Agitation Subscale, mean (SD)⁴⁴ Baseline: 3.13 (NR) vs. 3.87 (NR) End of treatment: 1.36 (NR) vs. 3.00 (NR) 4 week followup: 1.57 (NR) vs. 2.92 (NR)	Moderate	Direct	Imprecise	Consistent	Undetected	Low
	Patient General Behavior k=2; n=99	Behave-AD Global mean (SD)⁵⁰ Baseline: 0.9 (0.5) vs. 1.5 (0.7) vs. 1.3 (0.7) Post: 0.8 (0.4) vs. 0.7 (1.0) vs. 1.5 (0.8) 3 weeks follow-up: 1.1 (0.5) vs. 1.2 (0.6) vs. 2.2 (0.9) NPI:⁴⁴ results presented graphically; authors report lower scores post-intervention (F1,51=4.84, p<0.05); difference likely not significant at followup	Moderate	Indirect	Imprecise	Inconsistent	Undetected	Insufficient
Music vs. No treatment/Attention Control (for immediate reduction in agitation/aggression)	Patient Agitation/Aggression k=1; n=34	CMAI mean (SD)⁴⁵ Baseline: 18.41 (11.19) vs. 21.76 (9.09) Immediately post: 9.18 (11.11) vs. 21.88 (10.38) 10 min. post: 7.76 (9.55) vs. 20.88 (8.66) 20 min. post: 3.06 (5.44) vs. 20.47 (10.90)	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Music vs. interactive control	Patient Agitation/Aggression k=4; n=218	Behave-AD Aggressiveness, mean (SD)⁵⁰ Baseline: 1.5 (1.8) vs. 2.5 (2.4) Post-intervention: 1.5 (0.9) vs. 0.7 (1.0) 3 weeks followup: 1.3 (2.0) vs. 2.5 (2.2) *CMAI⁶¹ means– shown in figures; adjusted mean difference NS(F=2.89; p=0.09) CMAI,* mean (95% CI)¹¹ Baseline: 1.66 (1.42-1.91) vs. 1.54 (1.32-1.77) After first arm:1.67 (1.49-1.85) vs. 1.66 (1.37-1.96) Post crossover:1.65 (1.38-1.91) vs. 1.70 (1.44-1.97) *CMAI-SF, mean* (SD)⁴⁵ Baseline: 18.41 (11.19) vs. 16.47 (9.94) vs. 22.00 (11.94) Imm post: 9.18 (11.11) vs. 10.35 (11.20) vs. 8.59 (7.87) 10 min post: 7.76 (9.55) vs. 7.76 (9.55) vs. 7.06 (7.08) 20 min post: 3.06 (5.44) vs. 3.06 (5.44) vs. 3.76 (4.40)	Low to Moderate	Direct	Imprecise	Consistent	Undetected	Low
Music vs. interactive control	Patient General Behavior k=1; n=26	Behave-AD Global, mean (SD)⁵⁰ Baseline: 0.9 (0.5) vs. 1.5 (0.7) Post-intervention: 0.8 (0.4) vs. 0.7 (1.0) 3 weeks followup: 1.1 (0.5) vs. 1.2 (0.6)	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Aromatherapy (lavender) vs. no treatment/attention control	Patient Agitation/Aggression k=3; n=245	CMAI, mean (SD)⁶⁴ Posttreatment: 41.08 (8.24) vs. 41.72 (5.08) 3 week followup: 39.80 (7.27) vs. 42.13 (5.53) CMAI – aggressive behaviors¹⁴ No overall results reported; no statistically significant difference between groups on individual behaviors reported. C-CMAI, mean (SD)³⁴ Baseline: 63.17 (17.81) vs. 63.94 (SD 17.67) Post: 58.77 (16.74) vs. 63.90 (17.73)	Moderate	Direct	Imprecise	Consistent	Undetected	Low
Aromatherapy (lavender) vs. no treatment/attention control	Patient General Behavior k=2; n=98	NPI, mean (SD)¹⁵ Baseline: 31 (10) vs. 32 (11) 4 weeks: 18 (12) vs. 27 (12) CNPI, mean (SD)³⁴ Baseline: 24.68 (10.54) vs. 24.33 (10.08) Post: 17.77 (7.52) vs. 24.41 (10.24)	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Aromatherapy (melissa) vs. no treatment/attention control	Neuroleptic Use k=1; n=72	Prescribed additional psychotropic drugs during the study:⁴ 6% vs. 8% (SDs not reported)	Moderate	Indirect	Unclear	Unknown	Undetected	Insufficient
Aromatherapy (melissa) vs. no treatment/attention control	Patient Agitation/Aggression k=1; n=72	CMAI⁴ Proportion making 30% decrease in score: (60% vs. 14%, χ2=16.3; p<.0001) CMAI, median change⁴ -22.0 vs. -6.5 Z=4.1; p<.0001	Moderate	Direct	Unclear	Unknown	Undetected	Insufficient
Aromatherapy (melissa) vs. active control	Patient Agitation/Aggression k=1; n=77	PAS⁶⁵ median (95% CI) change from baseline 12 week followup: -0.7 (-1.7, 0) vs. -0.7(-1.7, 0)	Moderate	Direct	Unclear	Unknown	Undetected	Insufficient
	Patient General Behavior k=1; n=77	NPI, mean (95% CI) change from baseline 12 week followup: : –7.2 (–12.6, –1.7) vs. –10.0 (–17.2, –3.0)	Moderate	Direct	Unclear	Unknown	Undetected	Insufficient
	Patient Quality of Life k=1; n=77	Blau QoL, mean (95% CI) change from baseline 12 week followup: 17 (–13, 47) vs. –2 (–34, 30)	Moderate	Indirect	Unclear	Unknown	Undetected	Insufficient
Bright Light vs. no treatment/attention control k=4; n=225	Patient Agitation/Aggression k=4; n=225	Standardized Mean Difference, 95% CI¹^,⁶ 0.09 (-0.32 to 0.50) NPI Agitation/aggression, mean (SD)¹³ Morning bright light vs. evening bright light vs. standard light Baseline: 5.3 (3.5) vs. 3.7 (2.4) vs. 5.8 (3.4) Post-intervention mean: 5.5 (3.3) vs. 4.8 (2.6) vs. 4.3 (2.5) *Agitation – Behave-AD Aggression subscale*³⁸ No significant differences, did not present data (p>0.05)	Low to Moderate	Direct	Imprecise	Consistent	Undetected	Low
Bright Light vs. no treatment/attention control k=4; n=225	Patient General Behavior k=3; n=133	Crichton Royal Behavior Rating, mean (SD)⁶ Baseline: 34.2 (6.5) vs. 35.6 (7.6) Week 4: 41.3 (2.9) vs. 42.8 (1.4) Week 8: 43.8 (3.4) vs. 44.2 (2.5) MOUSEPAD, mean (SD)⁶ Baseline: 13.5 (11.6) vs. 13.4 (8.8) Week 4: 7.8 (7.9) vs. 7.8 (SD 4.3) Week 8: 8.0 (7.8) vs. 7.7 (3.7) NPI, mean (SD)¹³ Baseline: 29.4 (20.7) vs. 27.0 (15.7) vs. 24.1 (15.8) Post-intervention: 26.3 (13.9) vs. 27.5 (16.5) vs. 19.6 (10.8) Behave-AD, mean (SD)³⁸ Baseline: 14.9 (3.83) vs. 13.7 (3.49) Week 4: 12.6 (SD 4.79) vs. 10.7 (4.85)	Low to Moderate	Indirect	Imprecise	Consistent	Undetected	Insufficient
Therapeutic Touch vs. no treatment/attention control	Patient Agitation/Aggression k=1; n=51	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Therapeutic Touch vs. no treatment/attention control	Patient General Behavior k=2; n=108	See Report Text Table 4	Moderate	Indirect	Imprecise	Inconsistent	Undetected	Insufficient
Massage therapy vs. no treatment/attention control K=2; n=105	Patient Agitation/Aggression k=1; n=34	Baseline: 16.47 (9.94) vs. 21.76 (SD 9.09) Post:10.35 (SD 11.20) vs. 21.88 (SD 10.38) 10 min. post: 7.76 (SD 9.55) vs. 20.88 (SD 8.66) 20 min. post: 3.06 (SD 5.44) vs. 20.47 (SD 10.90)	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
	Patient General Behavior k=1; n=71	Behavior alterations improvement 3 months: 34/36 vs. 0/35 5 months: 28/35 vs. 32/36	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Massage therapy vs. interactive control	Patient Agitation/Aggression k=1; n=55	CMAI⁶⁶ mean (SD) Posttreatment Total: 27.76 (9.63) vs. 36.07 (9.72) Physical non-aggression: 10.08 (5.01) vs. 12.25 (4.52) Physical aggression: 5.36 (3.07) vs. 6.43 (3.50) Verbal nonaggression: 6.40 (3.44) vs. 9.57 (3.82) Verbal aggression: 5.92 (2.81) vs. 7.82 (3.76)	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Tailored Activities vs. Nontailored Activities k=4; n=334	Patient Agitation/Aggression k=4; n=334	Mean (SD)⁵⁹ Baseline: 16.7 (9.9) vs. 17.1 (9.8) During intervention: 8.4 (9.9) vs.10.0 (10.4) After intervention: 17.6 (10.3) vs. 17.0 (9.4) ABMI, mean (SD)⁹ Baseline: 8.76 (5.61)vs. 7.16 (7.61) Post: 2.08 (2.68) vs. 7.92 (9.09) Visual Analog Scale (0 to 100 based upon observation), mean (SD)³⁰ Baseline: 38.97 (20.54) vs. 32.59 (21.66) Posttest mean (SD): 30.54 (15.31) vs. 32.25 (20.16) (Pretest to Posttest * group: F_1,69=4.26; p=0.43) Nonverbal Behavior Observations – Aggression⁶⁰ mean (SE) Posttreatment: 0.016 (0.04) vs. 0.117 (0.04)	Moderate	Direct	Imprecise	Inconsistent	Undetected	Insufficient
Tailored Activities vs. Nontailored Activities k=4; n=334	Patient General Behavior⁶⁰ k=1; n=87	Nonverbal Behavior Observations – General restlessness, mean (SE) Posttreatment: 6.50 (5.66) vs. 5.28 (5.56)	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Tailored Activities vs. Tailored Activities K=2; n=158	Patient Agitation/Aggression k=2; n=158	CMAI, Least Square means (95%CI)²⁹ Baseline: 1.62 (0.9-2.4) vs. 2.46 (1.7-3.2) vs. 1.86 (1.1-2.6) vs. 1.88 (1.1-2.6) Post: 1.2 (0.3-2.0) vs.1.7 (0.9-2.5) vs.1.5 (0.6-2.3) vs.1.10 (0.3-1.9) CMAI, mean (CI)²⁸ Baseline: 2.85 (2.0-3.7) vs. 2.85 (2.0-3.7) vs. 2.85 (2.0-3.7) Post: 1.35 (0.5-2.2) vs. 1.09 (0.3-1.9) vs. 1.14 (0.2-4.0)	Moderate	Direct	Imprecise	Unknown	Undetected	Low
Aroma-acupressure vs. no treatment/attention control	Patient Agitation/Aggression k=1; n=113	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Exercise vs. interactive control k=1; n=170⁵⁷	Patient General Behavior k=1; n=170	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Multisensory room + massage + exercise vs. no treatment/attention control k=1; n=39²³	General Patient Behavior k=1; n=39	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
	Patient Distress, QoL k=1; n=39	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Multisensory room vs. no treatment/attention control k=1; n=32⁶⁷	Patient Agitation/Aggression k=1; n=32	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
	Patient General Behavior k=1; n=32	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Multisensory room vs. interactive control k=1; n=32⁶⁷	Patient Agitation/Aggression k=1; n=32	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Multisensory room vs. interactive control k=1; n=32⁶⁷	Patient General Behavior k=1; n=32	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Humor therapy vs. no treatment/attention control k=1; n=398³⁷	Patient Agitation/Aggression k=1; n=398	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
	Patient General Behavior k=1; n=398	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
	Patient Distress, QoL k=1; n=398	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Acupuncture k=1; n=76	Patient General Behavior k=1; n=76	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Massage vs. Ear Acupuncture k=1; n=75	Patient General Behavior k=1; n=75	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Acupressure K=1; n=133	Patient Agitation/Aggression k=1; n=133	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Structured Activities K=1; n=133	Patient Agitation/Aggression k=1; n=133	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Acupressure vs. Structured Activities K=1; n=133	Patient Agitation/Aggression k=1; n=133	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Reminiscence K=1; n=40	Patient General Behavior k=1; n=40	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Exercise K=1; n=134	Patient General Behavior k=1; n=40	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Exercise K=1; n=134	Adverse Effects	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Pleasant Experiences K=1; n=20	Patient General Behavior k=1; n=20	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Multisensory stimulation vs. Recreation K=1; n=40	Patient General Behavior k=1; n=40	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Activities of Daily Living vs. Psychosocial Activity k=1; n=127	Patient General Behavior k=1; n=127	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Simulated presence K=1; n=54	Patient Agitation/Aggression k=1; n=54	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Enhancing Family Visits k=1; n=66	Patient Agitation/Aggression k=1; n=66	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Enhancing Family Visits k=1; n=66	Patient General Behavior k=1; n=66	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient
Electro stimulation K=1; n=27	Patient Agitation/Aggression k=1; n=27	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Group Multistimulation vs. Leisure Activities k=1; n=40	Patient Agitation/Aggression k=1; n=40	See Report Text Table 4	Moderate	Direct	Imprecise	Unknown	Undetected	Insufficient
Group Multistimulation vs. Leisure Activities k=1; n=40	Patient General Behavior k=1; n=40	See Report Text Table 4	Moderate	Indirect	Imprecise	Unknown	Undetected	Insufficient