Impact of Fall-Related Behaviors as Risk Factors for Falls Among the Elderly Patients With Dementia in a Geriatric Facility in Japan

Abstract

The purpose of this study was to clarify potential fall-related behaviors as fall risk factors that may predict the potential for falls among the elderly patients with dementia at a geriatric facility in Japan. This study was conducted from April 2008 to May 2009. A baseline study was conducted in April 2008 to evaluate Mini-Mental State Examination, Physical Self-Maintenance Scale, fall-related behaviors, and other factors. For statistical analysis, paired t test and logistic analysis were used to compare each item between fallers and nonfallers. A total of 135 participants were followed up for 1 year; 50 participants (37.04%) fell during that period. Results of multiple logistic regression analysis showed that the total score for fall-related behaviors was significantly related to falls. It was suggested that 11 fall-related behaviors may be effective indicators to predict falls among the elderly patients with dementia.

Introduction

As the number of elderly population increases in Japan, so does the number of elderly patients with dementia. It is estimated that there were 2 410 000 elderly patients with dementia in 2011, and this number will reach its peak at 3 550 000 in 2036. ¹ Dementia and cognitive impairment are well-known major risk factors for falls and fall-related injuries in the elderly patients. ^2,3 Because the elderly patients with dementia are likely to have atypical symptoms such as spatial agnosia and apraxia, which may include neurological symptoms, gait disturbances, and balance concerns, the elderly patients with dementia are likely to experience repeated falls and injuries and may eventually become bedridden.

The use of physical restraints, which was part of the primary strategy to help prevent elderly patients with dementia from falling, has been prohibited at facilities since the year 2000 due to the implementation of Long-Term Care Insurance in Japan. However, it is reported that 40% of facilities are still using physical restraints for their patients to prevent problems such as complaints, lawsuits, and compensation damages for injuries related to a fall. ⁴ Currently, these intentionally neglected facility obligations are being reviewed pending a lawsuit and are being discussed among medical care professionals with regard to elderly patients with dementia. It is no exaggeration to say that the prevention of falls and injuries within the elderly population with dementia is an urgent matter as well as a steadfast challenge to overcome in the field of nursing.

In terms of fall prevention research reports in Western countries, several authors ^{5 –7} have conducted intervention studies using large-scale populations from multiple facilities and randomized control trials (RCTs); systematic reviews and meta-analyses have also been published. ^8,9 Chang et al, ⁵ who synthesized 40 RCTs regarding fall prevention interventions, reported that multifaceted fall risk assessments and a management program are effective for both reducing and preventing falls. Gillespie ⁸ conducted a meta-analysis based on 62 previously published research articles to determine the most efficient intervention to prevent falls. Results showed that a multilateral intervention within a wide range of specialties, including risk factor screening and an intervention program for the elderly patients staying at homes and facilities, is effective. Similarly, Cameron et al ⁹ pointed out that the following factors were efficient methods for fall prevention intervention at a care facility or a hospital: addressing the various risks associated with falling; developing an individualized intervention program for each patient; and implementing intensive long-term support.

Based on results of their 6-month study, in which they observed elderly patients with dementia staying at an elderly care facility, Eriksson et al ¹⁰ reported that 62% of participants fell during the study period. However, in terms of RCTs regarding fall prevention, there is no enough evidence for multifactorial interventions to be proven effective in meta-analyses, because of problems such as a lack of full description of the quality of care and the intervention methods used in different facilities. ¹¹ It was reported that Behavioral and Psychological Symptoms of Dementia (BPSD), ¹² which include variables such as irritation or wandering, can cause a fall—especially among the elderly patients with dementia. In addition, various elements such as gait or balance problems due to cranial neuropathy are related to dementia in its progressive state ^10,13 and may trigger falls.

It is difficult to analyze the actual cause of a fall because elderly patients with dementia often cannot remember how the incident occurred. However, many elderly patients with dementia who are institutionalized at geriatric facilities are at serious risk of falling due to unpredicted behaviors that contribute to a fall.

Because the elderly patients with dementia have several serious risks for falling, nurses are not able to predict falls among these patients even if they use a fall risk assessment tool. ¹⁴ However, nurses who assess fall-related behaviors such as agitation, resisting care, and wandering indicate that these behaviors can easily lead to a fall. ^15,16 Previous studies have not determined what kinds of fall-related behaviors lead to falls. The purpose of this study is to clarify fall-related behaviors as fall risk factors that may predict potential falls among the elderly patients with dementia at geriatric facilities in Japan.

Methods

Participants included 135 elderly patients (24 men and 111 women) residing at a geriatric facility from April 2008 to March 2009. The mean age of participants was 86.94 years (±7.40); there was a significant difference in ages between men and women (83.88 ± 9.84 vs 87.52 ± 6.66 years, respectively; P = .028 [chi-square test]). Diagnoses of study participants included Alzheimer’s disease (89), cerebral vascular dementia (36), dementia with Lewy bodies (5), frontotemporal dementia, and Pick’s disease (5).

This study was approved by the Screening Committee of Research Ethics at Hamamatsu University School of Medicine. Due to their cognitive impairment, many participants were unable to comprehend various issues associated with conducting a research study, such as the purpose of the study. Therefore, our researchers explained to them through their family (or via a deputy) the following points of concern: the aim of the study, privacy protection issues, and the method used for reporting results of the study. Written consent was received from family members of the elderly patients with dementia.

The following data were included as baseline data at the beginning of April 2008: Mini-Mental State Examination (MMSE) score, ¹⁷ the Physical Self-Maintenance Scale (PSMS) score, ¹⁸ Behavioral Symptoms in Alzheimer’s disease (Behave-AD) score, ¹⁹ the number of medications taken, and fall-related behaviors. The definition of a “fall” was floor contact with a part of the body other than the soles of the feet. ²⁰ From April 2008 to March 2009, nurses and caregivers were instructed to submit a patient accident report when they discovered a fall, regardless of whether an injury occurred.

Evaluation Methods

Mini-Mental State Examination

The MMSE ¹⁴ is an interview-based screening test consisting of 30 items on the following 6 subscales: orientation, registration, attention and calculation, recall, language, and visual constriction. Each item is scored from 0 to 5, with a total score ranging from 0 to 30. Higher MMSE scores represent better functioning. The MMSE was conducted at baseline (April 2008) before a 1-week observational period that took place between the fall of April 2008 to March 2009.

Physical Self-Maintenance Scale

The PSMS ¹⁸ consists of 1 to 5 rating scales that evaluate abilities to perform activities of daily living (ADL) among the elderly patients, including toileting, feeding, dressing, grooming, physical ambulation, and bathing. Scores range from 6 to 30 with higher scores indicating greater independence.

Behavioral symptoms in Alzheimer’s disease

The Behave-AD ¹⁹ evaluates the medicinal effects of drug therapy for Alzheimer’s disease based on the following 8 items: (a) paranoid and delusional ideation, 7 items (0-21); (b) hallucinations, 5 items (0-15); (c) activity disturbances, 3 items (0-9); (d) aggressiveness, 3 items (0-9); (e) diurnal rhythm disturbances, 1 item (0-3); (f) affective disturbances, 2 items (0-6); (g) anxieties and phobias, 4 items (0-12); and (h) global rating. Higher Behave-AD scores represent worse functioning.

Number of medications taken

The researchers asked in-charge nurses how many different kinds of medications each participant took daily.

Fall-related behaviors

First, based on previous studies regarding falls among elderly patient with dementia, we determined more than 40 items associated with BPSD and extracted behaviors related to falling. Next, we conducted focus group interviews with 6 nurse experts who had nursing experience with elderly patients with dementia to identify fall-related behaviors. We then extracted behaviors related to falling from the original 40 items based on whether the nurse experts considered the items to be related to falls or not (items were evaluated on a 5-point scale, where 5 = strongly agree and 1 = strongly disagree). Twenty items with a mean score of 3.5 or higher were extracted in this step. Finally, 6 nurse experts discussed these 20 items to evaluate whether they could be assessed by nurses during clinical practice (items were evaluated on a 5-point scale, where 5 = strongly agree and 1 = strongly disagree). The final list of fall-related behaviors included 17 items with a score of 3.5 or higher. The 17 fall-related behaviors included (1) engaging in unexpected sudden behavior; (2) being agitated and wandering; (3) experiencing delirium or changing consciousness; (4) experiencing unconsciousness or syncope; (5) resisting care from nurses and caregivers; (6) inability to maintain seated balance in a wheelchair; (7) experiencing Parkinson’s disease or similar symptoms, such as a forward-bent posture, brachybasia, and pulsion; (8) paying no attention to danger; (9) trying to transfer by oneself when using the toilet or walking in spite of an order not to; (10) dangerous behavior other than transferring as usual or when conditions are not favorable; (11) trying to leave bed during bedtime; (12) standing up abruptly from a wheelchair and attempting to walk; (13) wandering in a unit by oneself; (14) walking in an unbalanced or unsteady manner; (15) attempting to move, walk, stand, or transfer by oneself, although it is impossible to do so; (16) having a sudden desire to urinate or defecate; and (17) becoming agitated due to a desire to urinate or defecate.

The charge nurse who took care of the participants evaluated PSMS, Behave-AD, number of medications taken, and fall-related behaviors at baseline (April 2008) before a 1-week observational period that took place between the fall of April 2008 to March 2009.

Statistical Analysis

The t test, χ² test, and multiple logistic regressions were used to clarify behavioral risks of falls. All statistical analyses including t tests, χ² tests, and multiple logistic regression analyses were conducted using PASW statistics version 18.0 (IBM Corporation, Chicago, Illinois).

Results

A total of 135 participants (24 males and 111 females) were included in this study. There was a significant difference in mean ages between males (83.88 ± 9.84 years) and females (87.52 ± 6.66 years; P = .028, t test).

Table 1 describes the physical ambulation ability of participants based on the PSMS. Most participants (62.5% of males and 73.9% of females) were able to “ambulate with assistance.” Although results of gender differences of PSMS are not shown in Table 1, females (3.11 ± 0.71) showed significantly worse findings than males (2.71 ± 0.75) in terms of types of physical ambulation (P = .014, t test).

Table 1.

Description of Physical Ambulation Ability by Gender

Physical ambulation as assessed by the Physical Self-Maintenance Scale (PSMS)	Males		Females		Total
	Number	%	Number	%	Number	%
Goes about grounds or city	2	8.3	3	2.7	5	3.7
Ambulates within residence or about 1 block distance	5	20.8	7	6.3	12	8.9
Ambulates with assistance	15	62.5	82	73.9	97	71.9
Sits unsupported in chair or wheelchair	2	8.3	13	11.7	15	11.1
Bedridden more than half the time	0	0.0	6	5.4	6	4.4
Total	24	100	111	100	135	100.0

*t test, P = .014,

A total of 50 (37.04%) participants fell during the 1-year study. Among males, 6 of 24 (25.0%) fell and among females, 44 of 111 (39.6%) fell. Although more females fell than males, the difference was not significant (chi-square test, P = .245).

Table 2 shows the relationship between annual accidental falls, MMSE scores, and PSMS scores. Although the mean MMSE score of males, females, and overall participants who fell was not significantly lower than those who did not fall, the MMSE score in male participants who fell was lower than that of those participants who did not fall, although differences did not reach statistical significance. In the PSMS, the mean score for “feeding” was significantly lower among male participants who fell compared with male participants who did not fall. No significant differences were seen for females or overall participants. Similarly, no significant differences were seen between participants who fell and participants who did not fall for any other variable on the PSMS. Participants who fell tended to take more medications than participants who did not fall but differences did not reach statistical significance.

Table 2.

Relationship Between Annual Accidental Falls, Mini-Mental Examination (MMSE) Score, and Physical Self-Maintenance Scale (PSMS) Score

Item	Annual accidental falls	Males		t test (P value)	Females		T test (P value)	Total		t test (P value)
		Mean	SD		Mean	SD		Mean	SD
MMSE a	Fallers	13.33	9.99	.058	12.34	8.35	.685	12.46	8.46	.497
	Nonfallers	13.61	9.69		13.52	9.63		13.54	9.59
Physical Self-Maintenance Scale (PSMS) b
Toileting (1-5)	Fallers	4.17	1.33	.085	3.27	1.55	.850	3.38	1.54	.418
	Nonfallers	2.89	1.71		3.21	1.86		3.14	1.82
Feeding (1-5)	Fallers	1.00	0.00	.046	1.39	0.87	.159	1.34	0.82	.094
	Nonfallers	1.50	0.99		1.69	1.36		1.65	1.29
Dressing (1-5)	Fallers	3.00	1.27	.575	2.95	1.36	.462	2.96	1.34	.341
	Nonfallers	2.61	1.50		2.75	1.58		2.72	1.56
Grooming (1-5)	Fallers	3.17	1.72	.701	2.86	1.37	.263	2.90	1.40	.278
	Nonfallers	2.89	1.45		2.55	1.46		2.62	1.46
Physical ambulation (1-5)	Fallers	2.50	0.84	.445	2.98	0.40	.072	2.92	0.49	.104
	Nonfallers	2.78	0.73		3.19	0.84		3.11	0.83
Bathing (1-5)	Fallers	3.33	1.03	.500	3.30	1.03	.168	3.30	1.02	.113
	Nonfallers	3.00	1.03		3.00	1.14		3.00	1.11
Total PSMS (6-30)	Fallers	17.17	5.34	.586	16.75	5.37	.771	16.80	5.31	.591
	Nonfallers	15.67	5.87		16.39	6.97		16.24	6.73
Number of medications	Fallers	2.83	1.941	.407	3.61	2.082	.129	3.52	2.06
	Nonfallers	3.67	2.326		3.00	2.03		3.14	2.10	.308

Abbreviation: SD, standard deviation.

^a Lower scores represent greater cognitive impairment.

^b Higher scores indicate greater dependence.

A comparison of annual accidental falls and Behave-AD scores is shown in Table 3. In males, results comparing each item on the Behave-AD with annual accidental falls showed no significant differences among items. In contrast, in females, scores for paranoid and delusional ideation, activity disturbances, aggressiveness, affective disturbances, anxieties and phobias, and total Behave-AD scores were significantly higher among fallers than nonfallers. Among overall participants (both males and females), those who fell had significantly higher scores on the following Behave-D items: paranoid and delusional ideation, activity disturbances, aggressiveness, affective disturbances, anxieties and phobias, and total score.

Table 3.

Comparison of Annual Accidental Falls and Behavioral Symptoms in Alzheimer’s Disease (Behave-AD) Scores

Item	Males				t test (P value)	Females				t test (P value)	Total				t test (P value)
	Annual accidental falls					Annual accidental falls					Annual accidental falls
	Fallers		Nonfallers			Fallers		Nonfallers			Fallers		Nonfallers
	Mean	SD	Mean	SD		Mean	SD	Mean	SD		Mean	SD	Mean	SD
Paranoid and delusional ideation	1.000	2.449	0.556	1.042	.532	1.364	1.942	0.463	1.133	.003	1.32	1.98	0.48	1.11	.008
Hallucinations	0.500	1.225	0.556	0.984	.911	0.523	1.023	0.224	0.670	.065	0.52	1.03	0.29	0.75	.182
Activity disturbances	0.500	0.837	0.500	0.924	1.000	1.068	1.591	0.269	0.687	.003	1.00	1.53	0.32	0.74	.004
Aggressiveness	1.833	2.137	1.056	1.955	.418	1.318	2.197	0.463	1.259	.022	1.38	2.17	0.59	1.44	.025
Diurnal rhythm disturbance	0.667	0.817	0.278	0.575	.209	0.182	0.495	0.179	0.424	.976	0.24	0.56	0.20	0.46	.668
Affective disturbance	0.333	0.817	0.167	0.515	.560	0.546	0.926	0.194	0.468	.023	0.52	0.91	0.19	0.48	.019
Anxieties and phobias	0.333	0.817	0.167	0.515	.560	0.591	0.871	0.254	0.659	.032	0.56	0.86	0.24	0.63	.023
Total	5.167	6.998	3.278	4.309	.436	5.591	6.296	2.045	3.106	.001	5.54	6.31	2.31	3.41	.001

Abbreviation: SD, standard deviation.

Table 4 shows comparisons of annual accidental falls and the 17 fall-related behaviors by gender. Overall, there were significant differences between participants who fell and those who did not fall in terms of 11 of the original 17 items: No. 1, 2, 5, 6, 8, 9, 10, 12, 15, 16, and 17. There were no significant differences in any item between males who fell and males who did not fall. In contrast, females who fell showed significant differences from females who did not fall in terms of the following 11 items: No. 1, 2, 3, 5, 8, 9, 10,12, 14, 15, and 16.

Table 4.

A Comparison of Annual Accidental Falls and Fall-Related Behavior by Gender

		Males			χ2 test (P value)	Females			χ2 test (P value)	Total			χ2 test (P value)
		Annual accidental fall				Annual accidental fall				Annual accidental fall
		Fallers, n (%)	Nonfallers, n (%)	Total, n (%)		Fallers, n (%)	Nonfallers, n (%)	Total, n (%)		Fallers, n (%)	Nonfallers, n (%)	Total, n (%)
1. Engaging in unexpected sudden behavior	Yes	4 (44.40)	5 (55.60)	9 (100.00)	.150	17 (70.80)	7 (29.20)	24 (100.00)	.001	21 (63.64)	12 (36.36)	33 (100.00)	0.000
	No	2 (13.30)	13 (86.70)	15 (100.00)		27 (31.00)	60 (69.00)	87 (100.00)		29 (28.43)	73 (71.57)	102 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
2. Being agitated and wandering	Yes	2 (25.00)	6 (75.00)	8 (100.00)	1.000	10 (76.90)	3 (23.10)	13 (100.00)	.005	12 (57.14)	9 (42.86)	21 (100.00)	.035
	No	4 (25.00)	12 (75.00)	16 (100.00)		34 (34.70)	64 (65.30)	98 (100.00)		38 (33.33)	76 (66.67)	114 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
3. Experiencing delirium or changing consciousness	Yes	2 (28.60)	5 (71.40)	7 (100.00)	1.000	9 (69.20)	4 (30.80)	13 (100.00)	.032	11 (55.00)	9 (45.00)	20 (100.00)	.062
	No	4 (23.50)	13 (76.50)	17 (100.00)		35 (35.70)	63 (64.30)	98 (100.00)		39 (33.91)	76 (66.09)	115 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
4. Experiencing unconsciousness or syncope	Yes	0 (0.00)	1 (100.00)	1 (100.00)	1.000	3 (37.50)	5 (62.50)	8 (100.00)	1.000	3 (33.33)	6 (66.67)	9 (100.00)	.558
	No	6 (26.10)	17 (73.90)	23 (100.00)		41 (39.80)	62 (60.20)	103 (100.00)		47 (37.30)	79 (62.70)	126 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
5. Resisting care from nurses and caregivers	Yes	3 (42.90)	4 (57.10)	7 (100.00)	.307	13 (65.00)	7 (35.00)	20 (100.00)	.021	16 (59.26)	11 (40.74)	27 (100.00)	.008
	No	3 (17.60)	14 (82.40)	17 (100.00)		31 (34.10)	60 (65.90)	91 (100.00)		34 (31.48)	74 (68.52)	108 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
6. Inability to maintain seated balance in a wheelchair	Yes	3 (42.90)	4 (57.10)	7 (100.00)	.307	15 (53.60)	13 (46.40)	28 (100.00)	.117	18 (51.43)	17 (48.57)	35 (100.00)	.033
	No	3 (17.60)	14 (82.40)	17 (100.00)		29 (34.90)	54 (65.10)	83 (100.00)		32 (32.00)	68 (68.00)	100 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
7. Experiencing Parkinson’s disease or similar symptoms, such as a forward-bent posture, brachybasia, and pulsion	Yes	2 (28.60)	5 (71.40)	7 (100.00)	1.000	11 (50.00)	11 (50.00)	22 (100.00)	.332	13 (44.83)	16 (55.17)	29 (100.00)	.221
	No	4 (23.50)	13 (76.50)	17 (100.00)		33 (37.10)	56 (62.90)	89 (100.00)		37 (34.91)	69 (65.09)	106 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62..96)	135 (100.00)
8. Paying no attention to danger	Yes	4 (28.60)	10 (71.40)	14 (100.00)	1.000	24 (61.50)	15 (38.50)	39 (100.00)	.001	28 (52.83)	25 (47.17)	53 (100.00)	.002
	No	2 (20.00)	8 (80.00)	10 (100.00)		20 (27.80)	52 (72.20)	72 (100.00)		22 (26.83)	60 (73.17)	82 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
9. Trying to transfer by oneself when using the toilet or walking in spite of an order not to	Yes	4 (44.40)	5 (55.60)	9 (100.00)	.150	22 (57.90)	16 (42.10)	38 (100.00)	.007	26 (55.32)	21 (44.68)	47 (100.00)	.001
	No	2 (13.30)	13 (86.70)	15 (100.00)		22 (30.10)	51 (69.90)	73 (100.00)		24 (27.27)	64 (72.73)	88 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
10. Dangerous behavior other than transferring as usual or when conditions are not favorable	Yes	1 (14.30)	6 (85.70)	7 (100.00)	.629	12 (70.60)	5 (29.40)	17 (100.00)	.007	13 (54.17)	11 (45.83)	24 (100.00)	.048
	No	5 (29.40)	12 (70.60)	17 (100.00)		32 (34.00)	62 (66.00)	94 (100.00)		37 (33.33)	74 (66.67)	111 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
11. Trying to leave bed during bedtime	Yes	4 (33.30)	8 (66.70)	12 (100.00)	.640	11 (55.00)	9 (45.00)	20 (100.00)	.137	15 (46.88)	17 (53.13	32 (100.00)	.134
	No	2 (16.70)	10 (83.30)	12 (100.00)		33 (36.30)	58 (63.70)	91 (100.00)		35 (34.31)	68 (66.67)	103 (100.98)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
12. Standing up abruptly from a wheelchair and attempting to walk	Yes	3 (42.90)	4 (57.10)	7 (100.00)	.307	7 (87.50)	1 (12.50)	8 (100.00)	.006	10 (66.67)	5 (33.33)	15 (100.00)	.014
	No	3 (17.60)	14 (82.40)	17 (100.00)		37 (35.90)	66 (64.10	103 (100.00)		40 (39.22)	80 (78.43)	120 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
13. Wandering in a unit by oneself	Yes	1 (11.10)	8 (89.90)	9 (100.00)	.351	8 (42.10)	11 (57.90)	19 (100.00)	.803	9 (27.27)	19 (57.58)	28 (84.85)	.355
	No	5 (33.30)	10 (66.70)	15 (100.00)		36 (39.10)	56 (60.90	92 (100.00)		41 (40.20)	64 (64.71)	107 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
14. Walking in an unbalanced or unsteady oneself	Yes	1 (11.10)	8 (89.90)	9 (100.00)	.351	14 (63.60)	8 (36.40)	22 (100.00)	.015	15 (45.45)	16 (48.48)	31 (93.94)	.355
	No	5 (33.30)	10 (66.70)	15 (100.00)		30 (33.70)	59 (66.30	89 (100.00)		35 (34.31)	69 (67.65)	104 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
15. Attempting to move, walk, stand, or transfer by oneself, although it is impossible to do so	Yes	3 (33.30)	6 (66.70)	9 (100.00)	.635	18 (75.00)	6 (25.00)	24 (100.00)	.000	21 (63.64)	12 (36.36)	33 (100.00)	.000
	No	3 (20.00)	12 (80.00)	15 (100.00)		26 (29.90)	61 (70.10)	87 (100.00)		29 (28.43)	73 (71.57)	102 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (39.60)	67 (60.40)	111 (100.00)		50 (37.04)	85 (62.96)	135 (100.00)
16. Having a sudden desire to urinate or defecate	Yes	2 (50.00)	2 (50.00)	4 (100.00)	.251	11 (73.30)	4 (26.70)	15 (100.00)	.009	13 (68.40)	6 (31.60)	19 (100.00)	.000
	No	4 (20.00)	16 (80.00)	20 (100.00)		33 (34.70)	62 (65.30)	95 (100.00)		37 (31.90)	79 (68.10)	116 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (40.00)	66 (60.00)	110 (100.00)		50 (37.00)	85 (63.00)	135 (100.00)
17. Becoming agitated due to a desire to urinate or defecate	Yes	2 (66.70)	1 (33.30)	3 (100.00)	.143	10 (58.80)	7 (41.20)	17 (100.00)	.106	12 (60.00)	8 (40.00)	20 (100.00)	.000
	No	4 (19.00)	17 (81.00)	21 (100.00)		34 (36.60)	59 (63.40)	93 (100.00)		38 (33.00)	77 (67.00)	115 (100.00)
	Total	6 (25.00)	18 (75.00)	24 (100.00)		44 (40.00)	66 (60.00)	110 (100.00)		50 (37.00)	85 (63.00)	135 (100.00)

Table 5 shows results of a multiple logistic regression model of variables used to predict falls using number of annual accidental falls as a dependent variable. The total score for fall-related behaviors was calculated as follows: if the answer to an item was “yes,” it was worth one point; the points were summed to give the total score. The range of total scores for fall-related behaviors was from 0 to 11 points. Age, gender, MMSE score, PSMS score, Behave-AD score, number of medications taken, and the total score for the 11 fall-related behaviors that were significantly related to falls among all participants (Table 5) were used as independent variables. The total score for fall-related behaviors and gender were significantly related to falls (Note: First paragraph about males who fell and feeding difficulties—the low scores mean that they had lesser abilities to feed themselves, according to “Methods” “PSMS”).

Table 5.

Multiple Logistic Regression Model Using Number of Annual Accidental Falls as a Dependent Variable

	β	SD	P value	Exp (B)	RR	95% CI
Age	−0.024	0.028	.393	0.976	0.923	1.032
Gender a	1.398	0.615	.023	4.047	1.213	13.495
Mini-Mental Examination (MMSE)	−0.001	0.026	.965	0.999	0.949	1.051
Physical Self-Maintenance Scale (PSMS) score	−0.034	0.152	.822	0.966	0.717	1.303
Behavioral Symptoms in Alzheimer’s Disease (Behave-AD) Score	0.035	0.057	.545	1.035	0.925	1.158
Number of medications	0.149	0.107	.161	1.161	0.942	1.431
Total score for fall-related behaviors (11 items)	0.302	0.103	.003	1.353	1.106	1.655

Abbreviations: RR, relative risk; CI, confidence interval.

^a Gender: male = 1 and female = 2.

Discussion

Fall-related behaviors represent dangerous actions taken by the elderly patients with dementia and can be caused by core symptoms of dementia or BPSD. Thus, nursing interventions, including analyzing and reducing dangerous behaviors, are necessary. It is important to develop a strategy for predicting falls not only by evaluating physical and cognitive abilities but also by assessing behaviors. According to the relationship between PSMS scores and falls in males, male participants who fell had significantly lesser abilities to feed themselves than those who did not fall, which means that male participants who fell had higher abilities in terms of eating. However, males who fell had impaired ability for other items related to ADL, with no significant differences between fallers and nonfallers. Similarly, there were no significant differences in females between PSMS scores and falls.

The elderly patients with serious impairments, including those who are bedridden, generally cannot move on their own and thus have a lower likelihood of experiencing a fall. Buchner and Larson ²¹ pointed out the significant relationship between falling and wandering among the elderly patients with dementia. In this study, it was suggested that the elderly patients with dementia sometimes needed assistance with ADL, especially minor and moderate working and moving assistance.

Studies ^2,22 have already reported that BPSD such as agitation and wandering among the elderly patients with dementia can increase the risk of falling. In this study, although significant relationships between falling and Behave-AD items were not seen in male participants, they were identified among females and overall participants. In the overall participants, scores for paranoid and delusional ideation, activity disturbances, affective disturbances, anxieties and phobias, and total score were significantly higher among participants who fell than among participants who did not fall. It is possible that significant differences were not seen in the male population due to the small sample size (n = 24).

Wandering behavior, which is included as “C. Activity Disturbances” in the Behave-AD, may increase the risk of falls. Buchner and Larson ²¹ reported that the elderly patients with Alzheimer’s disease have 3 times greater risk of bone fracture compared with an age-matched sample without Alzheimer’s disease and 5 times greater risk if they demonstrate wandering behavior. These patients often lose their balance and fall while they are wandering. Therefore, it may be possible to reduce the prevalence of falls by preventing wandering behavior.

In addition, patients who fell in this study had higher scores on the Behave-AD in terms of aggressiveness, affective disturbances, and anxieties and phobias. Studies ^{23 –25} have reported that among the elderly patients with dementia, a symptom such as agitation can develop due to treatments such as use of physical restraints or certain medications. Hence, it is important to consider reducing agitation by administering alternative treatment options or providing proper nursing care. The so-called BPSD symptoms, such as agitation and wandering, tend to cause falls and also create obstacles to efforts to provide safe nursing care. However, labeling all fall-related behaviors among the elderly patients with dementia as “agitation” or “wandering” may not be appropriate as it can inhibit predicting other risk factors for falls among this population. Therefore, this study aimed to identify fall-related behaviors among the elderly patients with dementia and predict falls based on these behaviors. When comparing fall-related behaviors with the actual prevalence of falls, significantly more participants who fell had fall-related behaviors associated with 11 of 17 items. Behaviors such as delirium, consciousness disturbances, symptoms of Parkinson disease, leaving bed during bedtime, and wandering were not related to falls.

The elderly patients with delirium or consciousness disturbances have increased risk for falls because their attention can many times become scattered due to their altered level of consciousness. ² However, in our study, no significant relationships between these elements were identified. It is possible that the nurses who evaluated patients in this study did not possess enough knowledge about delirium among the elderly patients with dementia; therefore, further consideration regarding the methodology of this study is required. Also, a higher prevalence of falls existed among the elderly patients with dementia who were exhibiting elimination disturbances, such as urinary incontinence or disturbances in excretion behaviors. Excretion behaviors are related to privacy issues. The elderly patients with dementia who have impairments in ability and who clearly require assistance still feel a need to go to the restroom and perform this function independently, even if doing so is impossible. Hence, although some may have difficulty walking for various reasons, many still to go to a restroom by themselves, which tends to increase the risk of falls. In addition, some often lose balance or suffer from a change in their cardiovascular system when they transfer to a toilet seat, which can also cause falls. Additionally, it is also reported that some patients experience delirium or BPSD due to their frustration, both of which are related to their inability to express a concern with conditions of constipation or diarrhea. ²⁴ Hence, providing nursing care to assist with elimination behaviors among the elderly patients with dementia has the potential to aid in preventing falls.

In terms of gender, the mean age of females in this study was significantly higher than that of males. In addition, there were fewer males (n = 24) than females (n = 111). Although more females fell than males, the difference was not statistically significant. Impairments in physical ambulation are reported to be related to falls, ²⁶ and there were significant gender differences between males and females on the functions of physical ambulation in this study. Gambassi ²⁷ reported that females are older than men and are more likely to fall than men among patients with Alzheimer’s disease. In this study, females who had impairments in physical ambulation were more likely to fall. Moreover, there were no significant differences in any fall-related item between males who fell and males who did not fall. In contrast, females who fell showed significant differences from females who did not fall in terms of all 11 fall-related items. We suggest that females in this study might have been more affected by fall-related behaviors because they were older and had greater impairment of physical ambulation compared with males. Further studies with a larger number of participants are needed to clarify our findings.

Where risks caused by fall-related behaviors are concerned, our study showed a significant association between the prevalence of falls and performance of careless behaviors (ie, actions taken without exercising proper judgment). Additionally, the ability to identify various dangers within one’s surroundings tends to decrease among the elderly patients with dementia due to the negative influence of BPSD, spatial agnosia, and delirium, ^2,21 which can also cause falls.

As a result of this study, we could predict that falls are caused not only by gait disturbances but also by the state of behavior and where patients cannot identify environmental hazards due to abnormally increased activity, which in turn is caused by excessive emotional responses. Of the 17 fall-related items we included, we identified 11 items that showed a significant relationship with the prevalence of falls and then ran a multiple logistic regression model to determine which variables were associated with accidental falls. As a result, we found that the total score for these fall-related behaviors maintained a significant relationship with the prevalence of falls, even after controlling for variables such as age, gender, cognitive function, and number of medications. Hence, these 11 items were found to be effective in predicting falls among the elderly patients with dementia. We believe that it is possible to contrive a concrete structured fall prevention care plan for elderly patients with dementia using these 11 items together. Although Jensen ²⁸ evaluated the effectiveness of a multifactorial fall and injury prevention program such as residence-specific risk factors in elderly patients with cognitive impairment, the lower cognitive group did not respond well because Jensen ²⁸ did not target fall-related behaviors among the elderly patients with dementia. The fall-related behaviors identified in this study are more specific fall risk factors among patients with severe dementia, and the results of this study may be used to identify fall risk factors among the elderly patients with dementia and develop efficient methods to prevent falls. Further study is required to evaluate the validity and reliability of these 11 fall-related behaviors to determine whether they maybe used as an effective indicator with which to predict falls among the elderly patients with dementia.

Although dementia is an important risk factor for falling, ²⁶ nurses have not known how to assess behavior of the elderly patients with dementia. Based on the fall-related behaviors identified in this study, nurses may be able to assess specific fall-related behaviors to predict—and thus prevent—accidental falls among the elderly patients dementia in geriatric facilities.

One limitation of this study is that we evaluated fall-related behaviors only one time, at baseline. It is necessary to evaluate fall-related behaviors regularly to predict falling among the elderly patients with dementia living in a geriatric facility. In addition, this study included only 135 elderly residents. It is necessary to conduct larger studies to determine the effective predictors of fall-related behaviors.