Table 1.

Characteristics of included studies.

Aim	Method, sample size, setting	Characteristics of participant	Intervention	Measurement tools	Findings
1. Chan et al., 2021, China Hong Kong
Evaluate the feasibility and effects of a culturally adapted group-based Montessori Method for Dementia programme in the Chinese community.	RCT, n = 108 people with dementia, community centre, day care center and nursing home.	Age: 83.9 ± 7. Gender: females: 77; males: 31. Stages of dementia: mild to moderate. Average length of stay in home: not reported.	Montessori group: Memory Bingo or sorting pictures or words into categories, cutting and stringing beads, reading groups, on an individual or group basis. Control group: reading out newspapers, physical activities, and watching videos on a group basis.	1. Feasibility: attendance rate. 2. Engagement: The Menorah Park Engagement Scale. 3. Affects (mood): The Apparent Affect Rating Scale.	Feasibility: Overall attendance: 79.1%. Montessori group: 81.5%; Control Group: 76.3%, not significant between groups. Engagement: Montessori group showed higher constructive engagement in first 10 minutes (Wald chi-square = 15.2, p = 0.033), middle 10 minutes (Wald chi-square = 19.9, p = 0.006) and lower passive engagement in the last 10 minutes (Wald chi-square = 17.61, p = 0.014) of each session cross the entire session period than control group. No difference among the three settings. Affects: only pleasure and interest observed; anxiety/fear and sadness rarely observed. Montessori group showed a higher pleasure in the first 10 minutes (Wald chi-square = 25.4, p < 0.001) and middle 10 minutes (Wald chi-square = 25.7, p < 0.001) as well as interest in the first 10 minutes (Wald chi-square = 21.1, p = 0.004) and middle 10 minutes of each session (Wald chi-square = 19.1, p = 0.008). No difference among the three settings.
2. Chaudhry et al., 2020, Pakistan.
Evaluate feasibility and acceptability of a culturally adapted, group-based Montessori intervention.	Mixed-method, n = 12 dyads of people with dementia and care workers and researchers, nursing home.	Age: 68.3 ± 5.7. Gender: 3 females; 9 males. Stage of dementia: moderate. Average length of stay in home: not reported.	Montessori-based activities: identifying pictures of the famous building of Pakistan and the Pakistan version of the Flag puzzle.	1. Feasibility: recruitment rate and attrition rate. 2. Acceptability and tolerability of the intervention by participants. 3. Pre-post intervention evaluation: (1) Montreal Cognitive Assessment (MoCA); (2) Quality of Life Assessment in Dementia (DEMQOL); (3) Geriatric Depression Scale 15-item (GDS-15); (4) Cohen-Mansfield Agitation Inventory (CMAI); (5) Disability Assessment for Dementia (DAD).	Feasibility: 12 individuals with dementia were recruited and only one participant left, resulting in a retention rate of 83%. Acceptability and tolerability of the intervention: 89% of participants were feeling happy about the intervention and no adverse event happened indicating well tolerated by participants. Pre- and post-intervention evaluation: (1) total scores of the GDS-15 and CoMA no significant changes. However, a significant reduction in verbal/aggressive score of the CMAI subscale (p = 0.005). Slightly improved DAD on initiation domain (mean difference [MD] = 1.4, not significant) and performance domain (MD = 0.90, not significant) as well as slightly improved of DEMQOL total scores (MD = 5.45, not significant).
3. Gaspar and Westberg, 2020, USA.
Evaluate engagement and prescribed antipsychotic medications before and after Montessori-Inspired Lifestyle (MIL) implementation.	Quasi-experimental, n = 72 people with dementia, memory care units.	Age: 86.08 ± 7.37. Gender: 53 females; 19 males. Stage of dementia: mild to severe, most were moderate. Average length of stay in home: 21 months.	Montessori activities: reading group, cooking, music group, and unplanned activities, individual or group discussion activities.	Engagement: an engagement Observation Recording Form developed by Westberg et al., in 2017.	1. Engagement: significantly increased during mealtimes, planned activities and unplanned activities post-intervention (p < 0.05). 2. Negative correlations were found between length of exposure to MIL programme and minutes of positive engagement for meals (r2 = −0.356; p = 0.01) and unplanned activity (r2 = −0.27; p = 0.05), with planned activity (r2 = −0.316; p = 0.02), indicating that as time of expose to MIL increased, the minutes of positive engagement decreased. 3. Positive clinical changes: (1) number of participants not prescribed psychotropic medication (routine or PRN) increased from six to seven; (2) one participant discontinued PRN psychotropic medication; (3) number of participants on both routine and PRN psychotropic medication decreased from 11 to nine.
4. Giroux et al., 2010, Canada.
Evaluate short-term effects of the Montessori approach and activities.	A quasi-experimental design, n = 14, people with dementia nursing home.	Age: missing. Gender: missing. Stage of dementia: moderate to severe. Average length of stay in home: not reported.	Montessori group: Montessori activities: classifying objects and fitting shapes into holes, puzzles of maps, animals, or parts of plants. Placebo group: regular activities: music activities singing along, clapping hands or playing various musical instruments. Group games: casinos, quiz games, sandbags, bowling, etc, Inactivity group: participants were in their room, on the unit, or in a sitting area but not involved in any of the activities.	1. Affects: The Philadelphia Geriatric Center Affect Rating Scale. 2. Mood: The Dementia Mood-Pictures test. 3. Engagement: an observation scale developed by Kovach and Magliocco in 1998.	Gross affects score: Montessori group: 22.3 ± 1.1, Placebo group: 21.6 ± 1.9, Inactivity group: 19.3 ± 4.8, statistically different among these groups (p < 0.0001). Mood (positive): Montessori group: 10.1 ± 1.3, Placebo group: 10.5 ± 1.4, not significant different among different groups. Engagement score: Montessori group: 2.9 ± 0.8, Placebo group: 2.5 ± 0.5, statistically significant among groups (p < 0.0001). Intensity of the stimulation score: Montessori group: 1.8 ± 0.5, Placebo group: 2.3 ± 1.0, statistically significant difference in different groups (p = 0.0026). The length of time of active, passive and no participation: Montessori group: 92.4% of the time in active participation, 4.8% of the time in passive participation and 1.0% of the time in no participation at all. Placebo group: 53.1% of the time in activity participation, 41.8% in passive participation and 5.1% of the time not participating at all. Montessori group showed a significantly higher rate of active engagement than Placebo group (p < 0.0001). Correlation between different variables: a strong positive correlation was observed between global affects and engagement score (r = 0.52, p < 0.0001), more active engagement is positively associated with a more positive attitude. The intensity of stimulation had no correlation with engagement score.
5. Kao et al., 2016, Taiwan.
Compare long-term effectiveness of spaced retrieval (SR) training and SR training combined with Montessori activities (SR + M) for improving hyperphagic behaviours.	Mixed methods: RCT and interview, n = 140 people with dementia, dementia care units.	Age: 82.55 ± 5.95. Gender: females: 41; males: 99;. Stage of dementia: moderate. Average length of stay in homestay in home: 35.98 ± 32.43 months.	Memory training contents: remember where the food is placed, slow down eating speed and recognise the satisfaction message. SR group training: recall memory training contents correctly in six times intervals. SR + M group: the same procedure described as SR training and engaged with real action practising in Montessori-based activities (such as gently pressing, scooping) and cognitive training of matching and identification. Control group: routine activities, no memory training.	1. Hyperphagia behaviour: the scale of hyperphagia in individuals with dementia. 2. Pica behaviour: ‘Yes’ or ‘No’. 3. The changes in eating habits: self-developed observation form. 3. Short meal frequency: less than 10 min.	1. SR + M group showed the hyperphagic behaviour (β = −1.782, 95 CI% = [-3.20, −0.35], p < 0.05) and pica behaviour (β = −0.111, 95 CI% = [-0.20, −0.02], p < 0.05) continued to decrease for 3 months and short meal frequency reduction (β = −1.196, 95 CI% = [-2.09, −0.29], p < 0.01) lasted for 1 month after training. Changes in eating habit continued improve after 6 months but not significant. 2. SR + M group showed significant improvement in hyperphagic behaviour (β = −1.608, 95 CI% = [-3.13, −0.08], p < 0.05), short meal frequency (β = −1.094, 95 CI% = [-2.14, −0.04], p < 0.05) and change in eating behaviour (β = −0.287, 95 CI% = [-0.55, −0.01], p < 0.05) continued improvement after 6 months of training. Pica behaviour (β = −0.105, 95 CI% = [-0.20, −0.01], p < 0.05) continued improvement after 2 months of training. 3. Regarding to the hyperphagic behaviours, SR +M group showed significant reduction in intention to eat (β = −1.544, 95 CI% = [-2.63, −0.45], p < 0.01) and increased eating (β = −1.630, 95 CI% = [-3.06, −0.20], p < 0.05) in post-intervention. Significant reduction in rapid eating in post- training and the effect maintained 1 month after training (β = −1.215, 95 CI% = [-2.06, −0.36], p < 0.01). SR + M group showed rapid eating was significantly reduced after 6 months of training (β = −1.020, 95% CI = [-1.89, −0.14], p < 0.05), and continued reduction in increased eating after 3 months of training (β = −2.004, 95% CI = [-3.90, −0.09], p < 0.05). Regarding to the 6-month follow-up data of recall scores on MMSE, SR + M group had significant increase in post-training and 1 month (β = 0.334, 95% CI = [0.03, 0.63], p < 0.05), while SR + M group had significant increase in memory recall throughout the whole follow-up period (β = 0.304, 95% CI = [0.01, 0.60], p < 0.05).
6. Lin et al., 2010. Taiwan.
Compare the effectiveness of Space Retrieval (SR) and Montessori-based activities (M) in decreasing eating difficulty.	RCT, n = 81 people with dementia, dementia care units.	Age: 81 ± 6.37. Gender: females: 45; males: 36. Stages of dementia: mild to moderate. Average length of stay in home: 25.52 ± 19.34 months.	Space Retrieval group (SR): recall the 8-item eating procedures correctly at seven times intervals. Montessori group (M): mainly focused on eating abilities training: hand-eye coordination, scooping, pouring, squeezing; matching and differentiating of edible and non-edible items added to the programme. Control group: received and participated in the daily routine normally followed by their site’s schedule.	1. Feeding difficulty: the Chinese version of Edinburgh Feeding Evaluation in Dementia (EdFED). 2. Nutrition status: the Mini-Nutritional Assessment. 3. Body Mass Index (BMI). 4. Self-eating time: stopwatch. 5: Food consumption: the percentage of food consumed.	1. Both SR and M groups had a significant reduction in feeding difficulty (EdFED) score than control group post-intervention. SR group vs control group: Mean difference (MD) = −1.72, p < 0.05; Montessori group vs control group: MD = −1.54, p < 0.05. 2. Nutrition status (MNA): SR group showed significant increase post-intervention (MD = 3.64, p < 0.01) while M group showed significant decrease in nutrition post-intervention (MD = −2.58, p < 0.01). 3. Both SR group and M group showed no significant changes BMI and Body weight post-intervention. 4. Self-eating time: both SR group and M group showed a significant increase in self-eating time than control group post-intervention, p < 0.001. 5. Food consumption: M group showed significant reduction post-intervention (MD = −10.08, p < 0.05) while SR group showed slight increase but not significant.
7. Lin et al., 2011, Taiwan.
Evaluate the efficacy of Montessori intervention to improve the eating ability and nutritional status.	Experimental crossover design, n = 29 people with dementia, dementia care units.	Age: 82.90±5.96. Gender: females: 12; males: 17. Stage of dementia: most were moderate to severe. Average length of stay in home: 25.52 ± 19.16 months.	Montessori (M) group: focused on eating abilities training e.g., scooping food, sensory stimulation (play Mozart music), procedural movements (hand-eye coordination) and the activity leader’s ability to make conclusions and announcements for the next meeting. Control group: typical daily routine activities.	1. Feeding difficulty: the Chinese version of Edinburgh feeding evaluation in dementia (EdFED). 2. Eating behaviour: Eating behaviour scale. 3. Nutrition: The Chinese version of the Mini-nutritional assessment. 4. Body Mass Index (BMI). 5. Eating time: Stopwatch.	Feeding difficulty (EdFED): M group showed significant reduction of feeding difficulty score post-test (MD = −1.57 ± 3.41, p = 0.011), significant increase in physical assistance score (p = 0.039) and significant reduction fed by caregiver score (p = 0.040). Eating behaviour: both groups showed no significant changes post-test. M group showed slightly reduction in nutrition score post-test, not significant; but significant increase in self-eating frequency (p = 0.06) and self-eating time (p = 0.025). BMI in both groups, not significant.
8. Mbakile-Mahlanza et al., 2020, Australia.
Evaluate the impact of the Montessori activities (M) implemented by family members during nursing home visits.	RCT, n = 40 people with dementia, nursing home.	Age: 63.6 ± 10.8. Gender: females: 17; males: 3. Stages of dementia: mild to moderate. Average length of stay in home: 29.9 ± 21.7 months.	Montessori (M) group: 3-hour training session for family carer: 0.5-hour baseline assessment, 1-hour explaining Montessori activities in dementia, 1.5-hour brainstorming and practice possible activities based on the person’s ability and preference. Control group: 3-hour received education about dementia: half-hour baseline assessment, 1.5 hours for dementia education, and one hour in smaller groups to discuss the material just presented.	1. Affects: the Philadelphia Geriatric Center Affect Rating Scale. 2. Engagement: the Menorah Park Engagement Scale.	Affects: M group showed large effect size on neutral affects (Cohen’s d = 0.92), moderate effect size on pleasure (Cohen’s d = 0.69), interest (Cohen’s d = 0.51) and anger (Cohen’s d = 0.62), small effect size on contentment (Cohen’s d = 0.28) and anxiety (Cohen’s d = 0.40) but no effect on sadness. Engagement: M group showed large effect size on constructive engagement (Cohen’s d = 1.33) and passive engagement (Cohen’s d = 1.18), moderate effect size on non-engagement (Cohen’s d = 0.62) and self-engagement (Cohen’s d = 0.65).
9. Roberts et al., 2015, Australia.
Develop a composite model of care and evaluate its impact.	A pilot study, mixed-method, n = 16 people with dementia, memory care unit.	Age: 85 ± 4.1. Gender: 12 females; 4 males. Stage of dementia: moderate to severe. Average length of stay in home: not reported.	Household activities: feeding chickens, ironing, setting the table, sweeping the floor and folding clothing. Leisure activities: reading the newspaper and playing cards.	1. Medication usage: anti-psychotic or sedative medications. 2. Behavioural psychological symptoms of dementia (BPSD): the 29 Cohen Mansfield Agitation Inventory (CMAI).	1. Medication usage: substantial reductions in medication prescription. 4 (25%) individuals had no anti-psychotic or sedative medication prescribed at baseline and during the project.12 (75%), individuals had regular or PRN antipsychotic or sedative medication prescribed at baseline. 5 had a combination of both medications prescribed. At follow-up 6 months later, 7 of 12 individuals (58%) were no longer prescribed anti-psychotic or sedative medication. 9-months after baseline, an additional 2 individuals had ceased anti-psychotic medications. 18 months after baseline, no one was prescribed anti-psychotic medication and only two of the 12 individuals were still prescribed sedatives. 2. BPSD: A significant reduction in CMAI total scores from baseline (80.1 ± 14.3) to the 6-month follow-up (44.87 ± 13.64, p < 0.0001); a significant reduction in aggressive behaviour (p = 0.01) as well as the non-aggressive behaviour (p < 0.001) and verbal-agitation behaviour (p < 0.001).
10. Skrajner et al., 2012, USA.
(1) Determine if participants could effectively lead group activities; (2) Expand number of activities available to client leaders; and (3) Evaluate the amount of staff assistance needed by RAP leaders.	Pilot study, n = 11 people with dementia, nursing home.	Age: Montessori (M) group: 75.8 ± 15.5; Zgola (Z) group: 88.4 ± 6.2. Gender: 5 females; 1 male; Z group: five females. Stage of dementia: mild in M group and moderate in Z group. Average length of stay in home: not reported.	Two Montessori –based activities: Memories Squared (MS) and Reading Roundtable (RRT); and one Zgola-based activity: Meet and Remember (MR).	1. Number of players (participants) per session. 2. Number of times staff assisted during activity preparation. 3. Number of times staff needed to assist during activity or during discussion. 4. Number of times staff needed to assist activity conclusion. 5. Number of times staff needed to assist per activity overall.	1. MS activity: 10.5 (±2.7) players per session, 1.2 (±0.2) times staff assisted during preparation, 1.0 (±0.8) times staff assisted during gameplay, 0.6 (±0.5) times staff assisted during conclusion and 2.8 times assisted per game overall. 2. RRT activity: 6.6 (±2.4) players per session, 0.7 (±0.3) times staff assisted during preparation, 0.4 (±0.4) times staff assisted during the reading story, 0.5 (±0.6) times staff assisted during the discussion, 0.5 (±0.5) times staff assisted during conclusion and 2.1 times assisted per story overall. 3. MR activity: 4.9 (±1.7) players per session, 4.6 (±2.2) times staff assisted during preparation, 2.9 (±2.9) times staff assisted during the introductory discussion, 8.8 (±2.9) times staff assisted during the main discussion, 3.0 (±1.7) times staff assisted during conclusion and 19.3 times assisted per session overall.
11. Skrajner et al., 2014, USA.
(1) Evaluate the effects of activities led by clients; (2) Evaluate the effects of Montessori-based vs Zgola-based programming (ZBP); and (3) Evaluate the effects produced by researchers vs staff members.	Quasi-experimental study, n = 81, nursing home.	Phase 1: Montessori (M) group: n = 22; age: 83 ± 12; gender: 20 females; 2 males; Stage of dementia: moderate. Zgola (Z) group: n = 17, aged: 89 ± 5; gender: 17 females; stage of dementia: severe. Phase 2: M group: n = 22; age: 87 ± 6; gender: 16 females; 6 males; stage of dementia: moderate. Z group: n = 20, age: 88 ± 6; gender: 17 females; 3 males; stage of dementia: moderate. Average length of stay in home: not reported.	Phase 1: the activity leader (people with dementia) is trained by the researcher. Phase 2: activity leader trained by site staff trainer. Montessori (M) group: Memories Squared (MS) and Reading Roundtable (RRT). Zgola (Z) group: Meet and Remember (MR).	Engagement and affects: the Menorah Park Engagement Scale (MPES).	1. Engagement: constructive engagement (CE): phase 1: M group: significantly reduced ‘up to half’ CE (p < 0.05) but increased ‘more than half’ CE (p < 0.05) post-intervention. While in ZG, significantly increased in ‘up to half’ CE (p < 0.05) but significantly reduced ‘more than half’ CE (p < 0.01). Phase 2: M group showed significant reductions in ‘not at all’ CE (p < 0.01) but significantly increased ‘up to half’ (p < 0.01) and ‘more than half’ CE (p < 0.01). Z group showed significantly increased in ‘up to half’ CE (p < 0.01) but also a significant reduction in ‘more than half’ CE (p < 15). Passive engagement (PE): phase 1: M group showed significant reduction ‘more than half’ PE (p < 0.05) post-intervention. Z group showed significant reduction ‘up to half’ PE (p < 0.01) and significantly increased ‘more than half’ PE (p < 0.01). Phase 2: no significant changes in M group in terms of PE. Z group showed significant reduction ‘not at al’ PE (p < 0.05) but increased in ‘more than half’ PE (p < 0.01). Other engagement (OE): phase 1: significantly increased ‘not at all’ OE (p < 0.01) but reduced ‘up to half’ OE post-intervention (p < 0.01). Z group had no significant changes pre and post-interventions. Phase 2: M group showed significantly increased in ‘not at al’ OE (p < 0.01) while reduction in ‘more than half’ OE (p < 0.05). Similarly, Zgola group showed a significant reduction in ‘not at al’ OE (p < 0.05) but increased in ‘more than half’ OE (p < 0.05). Non-engagement (NE): phase 1: M group showed a significant decrease in ‘up to half’ NE (p < 0.05). 2. Affects: only pleasure is often observed. M group showed no significant changes pre and post-intervention in the two phases. While in Z group, phase 1 showed significant reduction ‘not at al’ pleasure (p < 0.05) and significantly increased ‘up to half’ pleasure (p < 0.05). In phase 2, Z group also showed significant reduction in ‘not at al’ pleasure (p < 0.01) and increased in ‘up to half’ pleasure (p < 0.01).
12. Wilks et al., 2019, USA.
Evaluate effects of Montessori-based activity programme.	Quasi-experimental with interrupted time series design, n = 43 people with dementia, nursing home.	Age: 86.3 ± 7.60. Gender: females: 32; males: 11. Stage of dementia: severe. Average length of stay in home: not reported.	Montessori activities: customized lesson planning, immediate feedback, intellectually stimulating exercises, sensory puzzles and games, group readings, and engagement in household activities.	1. Problem behaviours: 24-item Revised Memory + Behaviour Problems Checklist (RMBPC). 2. Social engagement: The six-item Index of Social Engagement. 3. Activities of daily living: The 20-item Bristol Activities of Daily Living Scale. 4. Quality of life: The Quality of Life-Alzheimer’s Disease Scale. 5. Anxiety: The Rating for Anxiety in Dementia (Shankar, Walker, Frost, & Orrell, 1999). 6. Psychological well-being: The Psychological Well-Being in Cognitively Impaired Person scale (PWB-CIP).	1. Problem behaviours: slightly reduced but no significance. 2. Social engagement: significantly decreased (Interval 1: 3.0 ± 2.19, interval 3: 2.6 ± 2.24, t = 2.197, p < 0.05). 3. Activity of daily living: significant reduction capacity of conducting activities of daily living (Interval 1: 1.2 ± 0.79, Interval 3: 1.0 ± 0.76, t = 2.984, p < 0.01). 4. Quality of life: slightly increased but NS. 5. Anxiety: slightly decreased but NS. 6. Psychological well-being significantly decreased (Interval 1: 42.5 ± 8.18, Interval 2: 41.2 ± 6.90, p < 0.01).
13. Wu et al., 2014, Taiwan.
Evaluate long-term effects of standardized and individualised spaced retrieval (SR) combined with Montessori-based activities on the eating ability.	Single-blind, quasi-experimental study with repeated measures, n = 90 people with dementia, nursing home.	Age: 82.8 ± 6.1. Gender: missing. Stage of dementia: mild to severe. Average length of stay in home: not reported.	Spaced Retrieval (SR) training contents: recall 8 items of SR eating procedures in seven interval trials. Montessori activities (M): mainly focused on eating abilities training: scoping, pouring, and squeezing. Standardised SR + M training group: 24 sessions. Individualised SR + M training group: depending on the cognition of the person, minimum of 10 sessions, mild dementia 23 sessions, moderate to severe dementia 35 sessions.	1. Eating difficulty: Chinese version of the Edinburgh Feeding Evaluation in Dementia. 2. Food consumption: the percentage of food consumed. 3. Body Weight: Body weight scale.	Eating difficulty score: both groups showed significantly long-terms effect in reducing eating difficulty scores: standardised group (MD: −0.39, p = 0.07), individualised group (MD: −0.30, p = 0.02). Food consumption: both groups showed significantly increased food consumption: standardised group (MD: 5.80, p < 0.001), individualised group (MD: 3.37, p < 0.007). Body weight: both groups showed significantly increased body weight: standardised group (MD: 0.99, p < 0.001), individualised group (MD: 0.99, p = 0.001). Overall, individualised group had a better result in food consumption and body weight than standardised group, however, standardised group showed a better outcome in eating difficulties than individualised group.
14. Wu and Lin, 2013, Taiwan.
Evaluate long-term effects of fixed/individualised spaced retrieval (SR) combined with Montessori-based activities (M) on nutritional status and body mass index and nutritional improvement’s moderating effect on depressive symptoms.	Single-blind, quasi-experimental study with repeated measures, n = 90 people with dementia, nursing home.	Age: 82.8 ± 6.1. Gender: missing. Stages of dementia: mild to severe. Average length of stay in home: not reported.	Spaced Retrieval (SR) training contents: to recall 8 items of SR eating procedures in seven interval trials. Montessori activities (M): mainly focused on eating abilities training: scoping, pouring, and squeezing. Standardised SR + M training group: 24 sessions. Individualised SR + M training group: depending on the cognition of the person, minimum of 10 sessions, mild dementia 23 sessions, moderate to severe dementia 35 sessions. Control group: routine care provided by the organisation.	1. Nutrition status: The Chinese version of Mini-Nutritional Assessment. 2. Body Mass Index (BMI): body weight (kg) divided by the square of height (m). 3. Depression: Chinese version of the Cornell Scale for Depression in Dementia (C-CSDD).	1. Nutrition status: both standardised group (slop increase: 0.84, p = 0.001) and individualised group (slop increase: 0.94, p = 0.001) showed significant improvement. The individualised group showed a medium effect size in improving nutrition status post-test (Cohen’s d = 0.75) and the effects lasted for 3-month follow-ups (Cohen’s d = 0.55). The individualised group showed a medium effect size in post-test (Cohen’s d = 0.67) and the effects lasted for 6-month follow-ups (Cohen’s d = 0.27). 2. BMI: the standardised group and individualised group with the slope increasing by 0.39 (p = 0.001) and 0.27 (p = 0.002). The standardised group showed no effects in post-test but a small effect size in 6-month follow-ups (Cohen’s d: 0.26). The individualised group showed no effects throughout the whole period. 3. Depression: only individualised group showed a significant reduction of depression score, with a slope decrease by 0.41 (p = 0.001). The standardised group showed medium effects in reducing depression scores in post-test (Cohen’s d: −0.50) and 1-month follow-up (ES: −0.76). The individualised group showed medium effects in post-test (Cohen’s d: −0.52) and a larger effect size in the 6-month follow-up (Cohen’s d: −0.73). Correlation: The changes in nutrition scores and the individualised intervention were the two important predictors of the changes in depression score (R2 = 0.44, SE = 1.21, F value: 20.99, p < 0.001), the improved nutrition score in individualised group was associated with a significant reduction in depression score (changes in nutrition score: B: −0.44, p < 0.001, individualised intervention: B = −0.32, p < 0.01). The individualised group produced significantly more changes in the nutrition score (B: 0.34, p < 0.01).
15. Yuen and Kwok, 2019. China, Hong Kong.
Evaluate effect of DementiAbility Methods: The Montessori Way (DMMW) on agitation.	RCT, n = 46 people with dementia, nursing home	Age: 86.17 ± 9.75. Gender: females: 36, Males: 10. Stages of dementia: moderate to severe. Average length of stay in home: not reported.	Montessori group: individualised activities participants' needs, interests, abilities, and skills. Control group: structured social activities: discussing newspapers and paly table games with standardized procedures.	Agitation: The Chinese Cohen-Mansfield agitation inventory (CMAI).	CMAI (frequency): Montessori group: Overall score significant reduced: pre: 45.61 ± 21.97; post: 35.52 ± 15.10, t = 3.39, p < 0.001, 95% CI = [3.92–16.25], Cohen’s d = 0.71. Verbal aggressive behaviour significant reduction: p < 0.05, Cohen’s d = 0.46. Physical nonaggressive behaviour significant reduction, p < 0.05, Cohen’s d = 0.58. Physical aggressive behaviour significant reduction: p < 0.001, Cohen’s d = 0.69. Aggressive and Physical nonaggressive behaviour scores slightly increased but NS. CMAI (disruptiveness): Montessori group: Overall score significant reduction: pre: 34.26 ± 12.62, post: 27.65 ± 7.43, t = 3.26, p < 0.001, 95% CI = [2.40–10.81], Cohen’s d = 0.68. Verbal aggressive score behaviour significant reduction: p < 0.05, Cohen’s d = 0.53. Physical nonaggressive score significant reduction: p < 0.05, Cohen’s d = 0.55. Physical aggressive behaviour score showed strong significant reduction: p < 0.001, Cohen’s d = 0.68.