Development and validation of the Fall Risk Assessment Scale for patients in rehabilitation hospitals: A methodological study

Abstract

Background and Aims

Falling is a serious threat for patient safety in hospitals. This study aimed to identify the risk factors of fall amongst rehabilitation patients and to use them for developing and validating the Persian version of Fall Risk Assessment Scale (FRAS).

Methods

The current methodological study was conducted in two phases. In the first phase, based on the review of the literature and investigation of the medical records of 251 patients selected via purposive sampling, the risk factors of fall were extracted and the FRAS was developed accordingly. In the second phase, the face and content validities of the designed scale were determined by cognitive interview and Content Validity Index (CVI) and to evaluate the construct validity, known‐groups comparison was performed. Its inter‐rater reliability was analyzed using the weighted Kappa Coefficient (κ*). The study adhered to COSMIN guidelines.

Results

Fall was significantly associated with disease diagnosis, used medications, history of fall, cognitive impairments, and three items of the Functional Independence Measure (toilet transfer, bed transfer, and shoer transfer). The CVI of the scale was 0.94. The risk for falls group had a significantly higher perceived fall risk than the no risk for falls group, thus establishing known‐group validity. Its weighted kappa coefficient was >0.85, its sensitivity was 73%, and its specificity was 82%.

Conclusion

The valid and reliable FRAS may accurately assess the level of Fall Risk patients in Rehabilitation wards, helping to predict fall during hospitalization. So, enabling the planning and implementation of effective caring interventions.

1. INTRODUCTION

Falling is a serious threat for patient safety in hospitals and is considered a major health problem in Inpatient Rehabilitation Hospitals (IRHs). ¹ , ² Defined as sudden, uncontrolled movements towards the ground, falls are a major health concern. ³ Patients in rehabilitation hospitals represent one of the most at risk populations for fall during hospitalization. ⁴ The previous studies revealed a higher risk of fall among patients in rehabilitation wards (between 10 and 17 falls per 1000 patient bed days) than in those hospitalized in acute hospital settings (3–6 falls per 1000 patient bed days). ³ , ⁵

The consequences of falling extend beyond physical harm to include psychological effects such as fear, reduced independence, and diminished physical activity. ⁶ Moreover, falls can disrupt rehabilitation programs, lengthen hospital stays, and complicate treatment for underlying disorders, imposing economic burdens on global healthcare systems. ² , ⁷ Recognizing the necessity for preventive strategies, particularly among at‐risk patients, becomes imperative. ⁴

A main step towards the prevention of fall amongst rehabilitation patients is the assessment of its risk factors as well as the identification of at‐risk patients. ² The previous studies demonstrated that the patients with lower scores in the Functional Independence Measure (FIM) as well as those admitted due to stroke, spinal cord injury (SCI), brain injury, and neurological disorders were at a higher risk of fall. ⁸ Generally, rehabilitation patients are evaluated in terms of FIM by an occupational therapist, physiotherapist, and speech therapist at admission. ⁴ Due to its strong correlation with the risk of fall, the FIM score can be utilized to assess at‐risk patients. ⁹ FIM is a multi‐dimensional instrument, which evaluates patients' functional independence in self‐care, motor, and cognitive dimensions. The construct and concurrent validity as well as the internal consistency (Cronbach's alpha: 93%–95%) of FIM have been approved previously. ¹⁰

There are other factors such as increased age, cognitive impairments, sleep disturbances, previous history of fall, taking such medications as tranquilizers, anticonvulsants, and antihypertensive drugs, length of hospital stay, urinary incontinence, visual or auditory impairment, vertigo, and poor balance, which have been previously taken into account in rehabilitation centers. ⁴ , ⁵ , ¹¹ , ¹² Such a sensitivity is important for identifying the risk factors and developing prevention strategies in this population. ¹³

At‐risk patients can be determined using fall risk assessment tools including the Morse Fall Scale, Heindrich II Fall Risk Model, St Thomas Risk Assessment Tool, and Downton Fall Risk Index. Although Montero‐Odasso et al. ¹⁴ recommend moving away from scored fall risk screening tools, standardized tools are essential for accurately identifying high‐risk patients upon hospital admission and enabling effective interventions. De Clercq et al. ¹⁵ highlight that, despite their limitations, these tools offer a structured approach to identifying key risk factors that might otherwise be overlooked. However, while most of these scales have been developed for both acute and rehabilitation hospitals, they may not be suitable for use in rehabilitation hospitals due to differences in the personal and environmental characteristics of the intended subjects. ⁴ , ¹⁶ The meta‐analysis by Yang et al. ¹⁷ reveals that frailty and fall risk are significantly influenced by various frailty assessment tools and contextual factors, such as economic development and geographic location. ¹⁸ These findings highlight the need for specialized risk assessment scales tailored to address the specific limitations and unique needs of patients in rehabilitation settings. ⁴ In general, at the time of admission to IRHs, patients are evaluated with respect to the risk of fall using Morse Fall Scale. ⁴ However, Rosario ¹⁹ reported that this scale was not able to correctly identify high‐risk patients among those who had or had not experienced falling in IRHs. Based on the results, 75%–90% of the patients were at a high risk of falling. Additionally, it showed a low risk of fall among high‐risk patients. For example, the patients who were not able to move following an SCI were considered low‐risk in this scale, while the incidence of fall has been reported to be 30%–75% in these patients. They might also fall while being transferred on wheelchairs. ¹⁸ , ²⁰ This has caused challenges for nurses and other clinical staff for evaluation of the risk of fall amongst rehabilitation patients. To address this issue, the Persian version of the Fall Risk Assessment Scale (FRAS) was developed, with established reliability and validity for patients admitted to rehabilitation hospitals, offering a tailored solution to the challenges posed by existing assessment tools in this specific healthcare setting.

2. MATERIALS AND METHODS

2.1. Study design, settings

This methodological study was conducted in Rofaydeh rehabilitation hospital, Tehran, Iran. This study was designed according to the Consensus‐based Standards for selection of health Measurement Instruments (COSMIN) guideline.

2.2. Instrument

In this study, an instrument was designed on the basis of CCFRAS, which was developed by Rosario et al. ⁴ in a rehabilitation hospital in the United States in 2013. This scale had a sensitivity of 90% for predicting fall amongst patients as well as an appropriate specificity for identifying 20%–30% of the at‐risk patients. ⁴ One of the limitations of CCFRAS is that it only assesses the disease diagnosis and FIM items, which are not comprehensive risk factors. Hence, the present study was carried out in two phases to identify the risk factors of fall amongst rehabilitation patients and to use them for developing the Persian version of FRAS and to establish its reliability and validity which can be applied in different setting with similar context.

Phase 1: Development of the FRAS.

In phase 1, the risk factors of falls were extracted from the review of the literature, previous hospital fall data and investigation of the medical records of 251 patients who were hospitalized in a rehabilitation hospital from January 2020 to December 2020. These documents were selected through purposive sampling and the patients who aged below 19 years and had incomplete FIM were excluded. Also relevant literature published since 1993 in English was searched on PubMed, CINAHL, Embase, using the key words “Accidental Falls,” “Fall Risk Assessment Tools,” “rehabilitation,” and “hospital.” Based on the medical records, and literature review, 25 subfactors related to fall risk were extracted. Then, an expert panel consisting of five members of the rehabilitation team (nurse, physician, rehabilitation specialist, and occupational therapist) was created to choose the risk factors to be included in the scale. Accordingly, age, sex, disease diagnosis, used medications, underlying diseases, history of fall, length of hospital stay, cognitive impairment, and 18 items of the FIM were selected, providing the ground for comparing fallers and non‐fallers. The patients were divided into a faller group that included the patients who had experienced a fall during hospitalization and a non‐faller group that included the patients who had not experienced a fall. It is worth mentioning that in case a fall occurs during the course of hospitalization in Rofaydeh rehabilitation hospital, it is recorded in a form by the staff and is reported to the patient safety department. Therefore, this department was asked to provide the research team with a report on the recorded falls during the study.

Phase 2: Reliability and validity evaluation of the FRAS.

2.3. Face validity

Cognitive interviewing is a technique used to evaluate the face validity and applicability of a survey or instrument. ²¹ Cognitive interviewing focuses on the questionnaire itself and not on the survey process. It is a method used to determine how respondents understand and answer questions and to assess which modifications are necessary to make questions easier to answer. ²² To identify problematic items, the face validity of the designed scale was assessed via cognitive interviews with 10 rehabilitation team member. In so doing, they were required to review the Persian version of the instrument in terms of ambiguity, difficulty, and relevance and to suggest words for substitution, if necessary. Accordingly, the necessary modifications were performed.

2.4. Content validity

The content validity of the designed scale was assessed according to the opinions of 10 experts based on Waltz and Bausell's Content Validity Index (CVI). ²³ To determine the CVI, the “cultural relevance“ criterion was used for each item via a 4‐point Likert scale with the following options: not relevant, ²⁴ somewhat relevant, ²⁴ relevant but needs modification, ²⁴ and completely relevant. ²⁴ Then, the items that had received the highest scores (3 or 4) were divided by the total number of experts. In this study, the CVI was 0.94 for the whole scale.

2.5. Known‐groups validity

In the designed scale, known‐groups validity was examined in terms of its ability to distinguish between subgroups of patients formed based on their previous status (no vs. yes). An independent sample t‐test was used to analyze differences in FRAS‐ Persian version scores between groups at risk and not at risk of falling.

2.6. Reliability

The reliability of the scale was assessed via inter‐rater agreement. The approach used to test inter‐rater reliability is the Kappa statistic, or the weighted Kappa for ordinal measures such as FRAS. The kappa value ranged from 0 to 1 with positive values indicating substantial agreement between the raters. Weighted kappa value of 0.63 indicated substantial inter‐rater reliability. In doing so, two raters were requested to simultaneously observe and evaluate the risk of fall among 30 patients in various wards of the rehabilitation hospital using the developed scale.

2.7. Data analysis

Considering descriptive statistics, quantitative variables were presented through mean ± standard deviation, ordinal variables and quantitative ones with skewed distribution were presented as median (interquartile range), and qualitative variables were reported as frequency (percentage).

Weighted kappa was used to assess the reliability of the scale. In addition, independent t‐test, Chi‐square, and Mann–Whitney tests were employed to identify the risk factors of fall and to determine the differences between fallers and non‐fallers. Considering the application of Chi‐square test for crosstabs with a small sample size, use was made of the significance level in Fisher's exact test. Moreover, to determine the odds ratio (OR) of the identified risk factors, binary logistic regression model was utilized that was adjusted for age, sex, and underlying diseases.

To compute the risk score, the relative risk (RR) was first calculated for all the factors. To simplify the interpretations, RR was multiplied by 10 and rounded to the nearest 5. As an example, 1.74 was rounded to 15 and 2.26 to 25. Then, the sum of the obtained scores was reported as the risk score. Furthermore, receiver operating characteristic (ROC) curve analysis was done to explore the sensitivity and specificity of the computed risk score and differentiation of the study groups on the basis of the risk score. Accordingly, the area under curve (AUC) was regarded as a criterion for the predictive power of actual groups; i.e., fallers and non‐fallers. Additionally, appropriate cut‐off points were determined for fallers and non‐fallers and their sensitivity and specificity were calculated. All analyzes were carried out using the IBM SPSS 25 and STATA 14 software and two‐sided p < 0.05 was considered statistically significant.

3. RESULTS

This study was conducted on 251 patients with the mean age of 49.01 ± 20.20 years. The average length of stay in rehabilitation wards was 22 days (interquartile range: 17–30). During hospitalization in rehabilitation wards, 64 cases of falling were recorded (25.5%). Among the patients, 32.3% had Cerebrovascular Accident (CVA), 0.8% had traumatic brain injury, 17.5% had SCI (C4‐T9), 29.5% had SCI (T10‐L5), and 12.7% suffered from neurological disorders. Besides, 10% of the patients had a history of falling and nearly 60% consumed anticonvulsants, antidepressants, or sleep medications. Considering underlying diseases, more than half of the patients had hypertension, diabetes, convulsion, or other disorders (Table 1).

Table 1.

The demographic characteristics of the patients admitted in the rehabilitation hospital.

Variable	Group	Frequency (percentage)
Age (years) Mean ± SD: 49.01 ± 2020	20–29	61 (24.3)
	30–39	44 (17.5)
	40–49	23 (9.2)
	≥50	123 (49.0)
Sex	Male	165 (65.7)
	Female	86 (34.3)
Disease diagnosis	CVA	81 (32.3)
	TBI	20 (8.0)
	SCI4	44 (17.5)
	SCI5	74 (29.5)
	Neurological disorders	32 (12.7)
History of fall	No	226 (90.0)
	Yes	25 (10.0)
Length of hospital stay (days) Median (Q1–Q3): 22 (17–30)	≤15	47 (18.7)
	16–21	75 (29.9)
	>21	129 (51.4)
Used medications	Anticonvulsants	14 (5.6)
	Tranquilizers/antidepressants	21 (8.4)
	Others	114 (45.4)
	No drugs	102 (40.6)
Cognitive impairment	Anxiety	31 (12.4)
	Depression	30 (12.0)
	Others	64 (25.5)
	No	126 (50.2)
Underlying diseases	Hypertension	67 (26.7)
	Diabetes mellitus	37 (14.7)
	Convulsion	10 (4.0)
	Others	30 (12.0)
	No	117 (46.6)

3.1. Identification of fall risk factors

In this study, the relationship between the probable risk factors and falling was assessed and the results have been presented in Table 2. Accordingly, fall was significantly associated with the history of fall, diagnosed disease, used medications, and cognitive impairments (p < 0.05). Based on the results, the proportion of patients diagnosed with CVA (45.3% in fallers and 27.9% in non‐fallers; p = 0.01) and neurological disorders (23.5% in fallers and 9.1% in non‐fallers; p = 0.005) was significantly higher among the individuals who had experienced falling compared to those who had not. However, the proportion of patients with SCI (C4‐T9) was significantly lower in fallers than in non‐fallers (7.8% vs. 20.9%; p = 0.02). The results also indicated a relationship between the history of fall and the incidence of fall during hospital stay. Accordingly, the individuals who had a history of fall experienced this event in the hospital, as well (27.9% in fallers vs. 3.2% in non‐fallers; p < 0.001). Furthermore, the participants experiencing a fall consumed significantly more sleep medications and antidepressants in comparison to the non‐faller group (p = 0.02). Considering the cognitive impairments, the prevalence of depression was three folds higher amongst fallers compared to non‐fallers (p = 0.002). Nonetheless, no significant difference was observed in the distribution of age, sex, length of hospital stay, and comorbidities in the two groups (p > 0.05).

Table 2.

Comparison of the fallers and non‐fallers regarding the probable risk factors.

Variable	Group	Fallers (64, 25.5%)	Non‐fallers (187, 74.5%)	p‐value*
Age (years)	20–29	15 (23.4)	46 (24.6)	0.62
	30–39	14 (21.9)	30 (16.0)
	40–49	4 (6.3)	19 (10.2)
	≥50	31 (48.4)	92 (49.2)
Sex	Male	37 (57.8)	128 (68.4)	0.13
	Female	27 (42.2)	59 (31.6)
Disease diagnosis	CVA	29 (45.3)	52 (27.9)	0.01
	TBI	4 (6.3)	16 (8.6)	0.61
	SCI (C4‐T9)	5 (7.8)	39 (20.9)	0.02
	SCI (T10‐L5)	20 (31.3)	54 (28.9)	0.75
	Neurological disorders	15 (23.4)	17 (9.1)	0.005
History of fall	Yes	19 (29.7)	6 (3.2)	<0.001
Length of hospital stay	≤15	8 (12.5)	39 (20.9)	0.11
	16–21	25 (39.1)	50 (26.7)
	>21	31 (48.4)	98 (52.4)
Used medications	Anticonvulsants	6 (9.4)	8 (4.3)	0.20
	Tranquilizers/antidepressants	10 (15.6)	11 (5.9)	0.02
	Others	32 (50.0)	82 (43.9)	0.47
Cognitive impairment	Anxiety	5 (7.8)	26 (13.9)	0.27
	Depression	15 (23.4)	15 (8.0)	0.002
	Others	15 (23.4)	49 (26.2)	0.74
Underlying diseases	Hypertension	20 (31.3)	47 (25.1)	0.41
	Diabetes mellitus	11 (17.2)	26 (13.9)	0.54
	Convulsion	5 (7.8)	5 (2.7)	0.13
	Others	7 (10.9)	23 (12.3)	0.82

^*Chi‐square test, wherever necessary, p‐values were reported based on the exact methods.

3.2. The relationship between the FIM score and fall

To determine the relationship between the FIM score and fall, fallers and non‐fallers were compared with respect to the median (Q1–Q3) of the FIM items (Table 3). Based on the results, three FIM items; i.e. bed transfer (p < 0.001), shower transfer (p < 0.001), and toilet transfer (p < 0.001), were significantly correlated to the rate of fall. Additionally, the fallers showed lower motor independence compared to the non‐fallers (Table 3).

Table 3.

Comparison of the fallers and non‐fallers regarding FIM.

Dimension	Variable	Fallers	Non‐fallers	p‐valuea
Self‐care	Eating	3 (1–5)	3 (1–6)	0.09
	Grooming	3.5 (2–5)	4 (2–7)	0.24
	Bathing	5 (4–6)	6 (4–7)	0.17
	Dressing‐upper	5 (3–6)	6 (3–7)	0.17
	Dressing‐lower	6 (5–7)	7 (5–7)	0.35
	Toileting	6 (5–7)	7 (5–7)	0.08
	Bladder management	5 (1–7)	5 (2–7)	0.39
	Bowel management	3 (1–7)	6 (1–7)	0.09
Mobility	Transfers (bed, chair, WC)	5 (3–6)	7 (5–7)	<0.001
	Transfers (toilet)	5 (4–7)	7 (6–7)	<0.001
	Transfers (tub, shower)	5 (3.3–7)	7 (5–7)	<0.001
	Walk/wheelchair	5 (4–7)	6 (4–7)	0.27
	Stairs	7 (7–7)	7 (7–7)	0.50
Cognitive	Comprehension	1 (1–1)	1 (1–1)	0.81
	Expression	1 (1–1)	1 (1–1)	0.89
	Social interaction	1 (1–1)	1 (1–2)	0.72
	Problem‐solving	1 (1–1)	1 (1–2)	0.38
	Memory	1 (1–1.8)	1 (1–3)	0.14

^a Mann–Whitney U‐test.

3.3. Multivariate analysis and derivation of the fall risk scores

To calculate the risk score, significant basic variables and FIM items were entered into the logistic regression model and OR and RR were computed for these risk factors. As mentioned earlier, this model was adjusted for age, sex, and comorbidities. In addition, the fall score was determined on the basis of RR. In this way, the RR was multiplied by 10 and rounded to the nearest 5. It should be noted that the risk score was calculated for the FIM items with 1 or 2 scores. Based on the results, the patients with 1 or 2 scores in toilet transfer, shower transfer, and bed transfer received the risk scores of 20, 25, and 25, respectively. Considering the other effective factors in falling, patients diagnosed with CVA, SCI (C4‐T9), and neurological disorders received the scores of 15, 25, and 20, respectively. Finally, 35, 20, and 25 scores were assigned to the patients who had a history of fall, consumed tranquilizers, sleep medications, or antidepressants, and suffered from depression, respectively (Table 4).

Table 4.

Computation of the risk score based on the logistic model.

Variable	Group	Fallers (64, 25.5%)	Non‐fallers (187, 74.5%)	ORa (95% CI)	p‐value	RR	Score
CVA	Yes	29	52	2.92 (1.31–6.51)	0.01	1.74	15
	No	35	135
SCI (C4‐T9)	Yes	59	148	3.03 (1.09–8.44)	0.03	2.45	25
	No	5	39
Neurological disorders	Yes	15	17	2.87 (1.27–6.49)	0.01	2.10	20
	No	49	170
Previous fall	Yes	19	6	16.9 (5.81–49.1)	<0.001	3.57	35
	No	49	181
Tranquilizers/antidepressants	Yes	10	11	3.49 (1.35–9.00)	0.01	2.03	20
	No	54	176
Depression impairment	Yes	15	15	3.24 (1.42–7.43)	0.005	2.26	25
	No	49	172
Bed transfer 1, 2	Yes	15	17	3.39 (1.52–7.58)	0.003	2.10	20
	No	49	170
Toilet transfer 1, 2	Yes	10	8	4.27 (1.56–11.7)	0.005	2.40	25
	No	54	179
Bath/shower transfer 1, 2	Yes	12	11	3.65 (1.48–9.03)	0.005	2.29	25
	No	52	176

Abbreviations: CI, confidence interval; OR, odds ratio; RR, relative risk.

^a Adjusted for age, sex, and comorbidity status.

3.4. Risk assessment scale sensitivity and specificity and determination of the appropriate cut‐off point

The computed risk score ranged from 0 to 210. In this study, the risk score ranged from 0 to 140, with the mean of 42.79 ± 29.71 and median of 40. The mean risk score was 34.44 ± 23.70 in the fallers group and 67.19 ± 32.14 in the non‐fallers group, and the difference between the two groups was statistically significant (p < 0.001). In addition, the AUC for determining the power of the risk score for classification of the participants into faller and non‐faller groups was 78% (Figure 1). This implied that using the calculated risk score, 78% of the individuals could be categorized into the faller and non‐faller groups.

Figure 1.

Area under the receiver operating curve for classifying the participants into the faller and non‐faller groups based on the computed risk score.

According to the analysis of sensitivity and specificity using ROC curve, a score ≥50 was considered the appropriate cut‐off point for the risk score, at which the best values were obtained for sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio. Considering the sensitivity of 0.73 and specificity of 0.82 for the cut‐off score ≥50, in case of obtaining scores ≥50, individuals are 73% likely to fall, while if they gain scores <50, they are 82% likely not to fall.

3.5. Known‐groups validity

To test known‐group validity, total fall risk scores were categorized into two groups by cut‐off score (50 points). The risk for falls group had a significantly higher perceived fall risk according to the FRAS‐Persian version than the no risk for falls group (p < 0.001), thus establishing known‐group validity (Table 5).

Table 5.

Known‐group comparison of CCFRAS scores between the fall risk for falls group and no risk for falls group (N = 251).

	Risk for falls (score ≥50)	No risk for falls (score <50)
Items	Mean ± SD	Mean ± SD a	p‐valueb
CVAa	52.1	29.31	<0.001
Neurological disorders	70 ± 31.34	38.81 ± 27.35	<0.001
Previous fall	91.60 ± 26.68	37.39 ± 24.72	<0.001
Tranquilizers/antidepressants	70.48 ± 30.53	40.26 ± 28.38	<0.001
Depression impairment	81.33 ± 27.75	37.56 ± 25.91	<0.001
Bed transfer 1, 2	69.74 ± 28.86	38.84 ± 27.76	<0.001
Toilet transfer 1, 2	82.22 ± 21.36	39.36 ± 27.44	<0.001

^a CVA: Cerebrovascular Accident; SD: standard deviation.

^b Independent samples'‐test or Mann–Whitney U‐test.

3.6. Reliability

The weighted kappa coefficient was found to be >0.85 for all the items, which revealed a proper agreement between the two raters.

4. DISCUSSION

Falls are preventable safety incidents that healthcare professionals must address as avoidable health issues rather than inevitable accidents. Accurately assessing patients' perceptions of their fall risk is a crucial step in fall prevention. ²⁵ This study identified preliminary fall risk factors through literature review and medical record analysis of rehabilitation hospital patients. Researchers then conducted construct and content validity testing to develop a scale for rehabilitation care professionals to identify patients at high risk of falling.

The study findings revealed a significant difference between the fallers and non‐fallers regarding neurological disorders, CVA, and SCI (C4‐T9). Additionally, the patients diagnosed with CVA showed the highest incidence of falling. These findings are consistent with previous research, in which, disease diagnosis was associated with a higher incidence of falls. ⁵ , ⁸ In this setting, stroke, traumatic brain injury, amputation, neurological disorders and SCI were mentioned as high‐risk medical diagnoses for fall. ⁴ , ¹³ , ²⁶ Additionally, the previous studies demonstrated that fall was highly prevalent among patients with MS ²⁷ and stroke ²⁸ and affected their daily lives. In the current research, neurological disorders, CVA, and SCI (C4‐T9) were entered into the risk assessment scale as high‐risk diagnoses.

The FRAS demonstrated, the results of logistic regression analysis revealed that the three FIM scores were correlated to an increased risk of fall. Accordingly, the fallers had lower scores of toilet transfer, bed transfer, and shower transfer. In prior investigations, FIM score at admission was found to be a risk factor of fall amongst rehabilitation patients. Those studies demonstrated that the total score of FIM as well as the score of its motor dimension were lower in fallers than in non‐fallers. ¹¹ , ²⁹ , ³⁰ In the research performed by Rosario et al., ⁴ , ³¹ bed transfer, toileting, shower transfer, and stairs showed lower scores as the risk factors of fall.

The present study findings revealed the consumption of sleep medications, tranquilizers, and anti‐depressants as one of the determinants of fall, which was in agreement with the results of other studies conducted on the issue. ³² , ³³

Cognitive impairments such as depression were another determinant of falling in the current investigation. In the same line, Rosario et al. ³¹ conducted a descriptive, retrospective study and came to the conclusion that depression and cognitive impairments were the predictors of negative outcomes in rehabilitation patients. Davoodi et al. ³⁴ also reported that the OR of fall was 8.2 folds higher in depressed older adults compared to those without depression.

History of fall was another risk factor in the present study. Rosario reported that patients with a history of previous falls experienced similar incidents again. ³¹ Other studies have also emphasized the history of fall as a determinant of this incident amongst patients in IRHs. ¹³ , ¹⁶ Unlike the findings of the present study, The Hendrich II Fall Risk Model (HIIFRM), which was designed to identify adult patients at risk of falling in acute care hospitals, does not consider a history of previous falls as a risk factor. ³⁵

A new risk assessment scale for falls in IRHs was developed, considering diagnoses, FIM items, medication consumption, fall history, and cognitive impairments as predictors. The scale demonstrated appropriate content validity (CVI) and showed significantly higher FRAS scores in patients at risk for falls compared to those not at risk, indicating its appropriateness for diagnosing falls in IRHs. However, further research is needed to establish its criterion validity using actual falls as the criterion.

The study found a high reliability for the FRAS, with a weighted kappa coefficient of >0.85 for all items. The main limitation was a small sample size due to reduced admissions during the COVID‐19 pandemic, and the exclusion of rare diagnoses. This limitation affected the construct validity, as confirmatory factor analysis was not performed. A larger follow‐up study with diverse patients is needed. Additionally, the retrospective nature of the study meant cognitive disorders were identified only through medical records, without additional diagnostic methods.

5. CONCLUSION

The FRAS developed in this study demonstrated satisfactory reliability and validity in a rehabilitation hospital setting. Therefore, it can help therapists identify the rehabilitation patients who are at risk of falling. Considering the study limitations, however, further studies with larger sample sizes are required to evaluate the scale's potential to identify high‐risk patients in various IRHs.

AUTHOR CONTRIBUTIONS

Shoeleh Rahimi: Conceptualization; data curation; formal analysis; investigation; methodology; writing—original draft; writing—review and editing. Hamid Reza Khankeh: Formal analysis; investigation; methodology; supervision; validation; writing—review and editing. Abbas Ebadi: Conceptualization; data curation; formal analysis; investigation; methodology; writing—review and editing. Batol Mohammadian: Data curation; formal analysis; writing—original draft; writing—review and editing. Narges Arsalani: Formal analysis; investigation; methodology; supervision; validation; writing—review and editing. Masoud Fallahi‐Khoshknab: Formal analysis; methodology; validation; writing—review and editing. Nazila Akbarfahimi: Validation; Writing—review and editing. Elham Loni: Validation; Writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

This study was a part of a PhD dissertation in nursing and all experimental protocols were approved by the Ethics Committee of the University of Social Welfare and Rehabilitation Sciences, Tehran, Iran (code: IR.USWR. REC.1399.058). The necessary permissions were obtained from the tool designer and the authorities of the Rofaydeh Hospital. Moreover, the nature and objectives of the research were completely explained to the participants. They were also informed about the voluntary nature of the study and their freedom to voluntarily participate in or withdraw from the study. The confidentiality of the patients' data was also observed throughout the research, and the results were published anonymously. Written informed consent was obtained from all participants.

TRANSPARENCY STATEMENT

The lead author Hamid Reza Khankeh affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Rahimi S, Khankeh HR, Ebadi A, et al. Development and validation of the Fall Risk Assessment Scale for patients in rehabilitation hospitals: a methodological study. Health Sci Rep. 2024;7:e70040. 10.1002/hsr2.70040

DATA AVAILABILITY STATEMENT

All authors have read and approved the final version of the manuscript. The data that support the findings of this study are available from the corresponding author upon reasonable request.