Falls are one of the most common complications of a stroke (Langhorne et al, 2000); approximately 5%–38% of patients fall at least once during their hospital stay (Nyberg and Gustafson, 1995; Teasell et al, 2002; Czernuszenko and Członkowska, 2009; Schmid et al, 2010; Persson et al, 2018). In particular, the fall rate among patients admitted to rehabilitation hospitals is high (Nyberg and Gustafson, 1995; Teasell et al, 2002; Czernuszenko and Członkowska, 2009). Falls can lead to adverse events such as fractures and head injuries (Schwendimann et al, 2008), prolonged hospitalization (Morello et al, 2015; Wong et al, 2016), and increased hospitalization costs (Morello et al, 2015). It is necessary to correctly identify the patients’ risk of fall and implement appropriate fall prevention measures.

Various risk factors for falls have been reported in patients with stroke admitted to rehabilitation hospitals, including the severity of motor impairment, impaired balance, cognitive function, urinary incontinence, number of medications, and previous fall history (Nyberg and Gustafson, 1997; Teasell et al, 2002; Nakagawa et al, 2008; Czernuszenko and Członkowska, 2009). Based on these findings, a number of screening assessment methods have been developed for the early detection of patients at high risk of falls (Nakagawa et al, 2008; Breisinger et al, 2014). Intuitive fall assessment by healthcare professionals, which is not limited to patients with stroke, has also been reported to have high fall prediction accuracy (Haines et al, 2007; Haines et al, 2009). A multicenter prospective study showed that physical therapists’ intuitive judgments of fall risk were effective in predicting falls in geriatric inpatients in a rehabilitation unit (Haines et al, 2009). A meta-analysis has also shown that nurses’ intuitive judgments of fall risk are equally good (Haines et al, 2007).

Fear of falling is associated with a risk of falling in patients with stroke and in older people (Jalayondeja et al, 2014; Freiberger et al, 2022). The Fall Efficacy Scale-International (FES-I) is a measure of patients’ own assessment of their fall risk (Yardley et al, 2005), and its score has been shown to be associated with fall occurrence in patients with stroke (Azad et al, 2014; Faria-Fortini et al, 2021). However, it has been pointed out that some patients may not be aware of their own risk of falls. In previous studies, hospitalized older patients with various diseases admitted to acute care hospitals (Lim et al, 2018) and outpatients with multiple sclerosis (Gunn et al, 2018) have reported discrepancies between their concern about fall risk and objective physiological fall risk. Similarly, in a study of patients with various diseases admitted to an acute care hospital, approximately half of the patients did not perceive themselves to be at high risk for falls, even when nurses assessed them as high risk (Twibell et al, 2015).

A difference in the perception of fall risk between patients and healthcare professionals may indicate that the patients themselves do not correctly perceive the risk of falls. In this situation, the risk of falls may increase due to misjudgments and actions beyond the acceptable range of one’s ability. However, to the best of our knowledge, no studies have examined the discrepancy in the perception of fall risk between patients with stroke admitted to rehabilitation hospitals and healthcare professionals. In addition, no studies have clarified the changes in the perception of fall risk over time or the relationship with falls.

This study aimed to determine the discrepancy in the perception of fall risk between therapists and patients with stroke admitted to a rehabilitation hospital, its change over time, and its association with the incidence of falls.

Among 655 consecutively enrolled patients with stroke, 426 met the inclusion and exclusion criteria and were analyzed (Figure 1).

Table 1 summarizes the characteristics of patients in each group at admission. The median (interquartile range) age of all participants (n = 426) was 71.0 (20.0) years, and the mean hospital stay was 79.0 (75.8) days. A total of 119 patients (27.9%) experienced one or more falls during hospitalization. Among them, 43 patients experienced multiple falls. The total number of falls was 200. The single- and multiple-fall groups had significantly longer hospital stays than the non-fall group.

Table 2 shows the clinical data at admission and discharge for each group. The severity of paralysis at both admission and discharge was significantly different between groups, with the multiple-fall group being the worst. The total FIM scores showed significant differences between the groups on admission, with the multiple-fall group showing the lowest values. At discharge, there were no significant differences between the fall groups.

Figure 2 shows the results of the FES-I for each group as assessed by the patients and physical therapists on admission and discharge. The median (interquartile range) FES-I score assessed by the patient and by physical therapists on admission was 27.0 (22.0) and 36.0 (32.5), respectively, for the non-fall group; 41.0 (27.3) and 55.0 (22.3), respectively, for the single-fall group; and 50.0 (25.5) and 61.0 (8.5), respectively, for the multiple-fall group. The FES-I assessed by patients and physical therapists at discharge was 20.0 (12.0) and 22.0 (16.0), respectively, in the non-fall group, 28.0 (19.3) and 34.0 (32.3), respectively, in the single-fall group, and 33.0 (19.5) and 47.0 (21.0), respectively, in the multiple-fall group. The FES-I scores assessed by patients as well as physical therapists improved significantly at discharge compared to that at admission. The FES-I scores assessed by patients were significantly lower than those assessed by physical therapists, indicating that patients perceived a lower risk of falls than physical therapists.

Figure 3 shows the discrepancy in perception of fall risk on admission and discharge for each group. Discrepancies in fall perception on admission and at discharge were median 3.0 (interquartile range 11.0) and 0.0 (6.0), respectively, in the non-fall group; 7.5 (16.3) and 3.0 (14.0), respectively, in the single-fall group; and 10.0 (21.0) and 9.0 (16.0), respectively, in the multiple-fall group. The discrepancy between the perception of fall risk between patients and physical therapists was significantly reduced from admission to discharge in the non-fall and single-fall groups, while the discrepancy was not reduced in the multiple-fall group. In the between-group comparisons of discrepancies in perception of fall risk, ANCOVA with number of days of hospitalization, age, and sex as covariates showed no significant group effects at admission (p = 0.092). There was a between-group effect at discharge (p = 0.011), and post hoc test results showed that the multiple-fall group was significantly larger than the non-fall and the single-fall group.

Table 3 shows the correlations between the total scores of the FES-I, SIAS-motor, and FIM as assessed by the patient and the physical therapist at admission and discharge. Each showed a significant relationship, with correlation coefficients higher for the results assessed by the physical therapist.

The intra-rater reliability (ICC 1.1) of the FES-I score rated by physical therapists was excellent at 0.997 (95% CI, 0.994 to 0.998; p < 0.001). Inter-rater reliability (ICC 2.1) between physical and occupational therapists was also good at 0.812 (95% CI, 0.673 to 0.896, p < 0.001).

The study revealed the difference in perceptions of fall risk between physical therapists and patients with stroke admitted to a rehabilitation hospital and its relationship to falls. At both admission and discharge, patients perceived their risk of falls to be lower than that perceived by physical therapists. Discrepancies in the perception of fall risk were significantly reduced at discharge compared to those at admission in the non-fall and single-fall groups, but not in the multiple-fall group.

Previous studies examining differences in fall risk between patients and healthcare workers have used different scales (Twibell et al, 2015; Gunn et al, 2018; Lim et al, 2018), which makes it difficult to show the extent of the differences in perception between patients and healthcare providers. In this study, we assessed the fall risk from the perspective of patients and healthcare workers using the same FES-I scale. Since the FES-I is a scale that originally reflects the psychological aspects of the patients themselves, we checked the reliability of other people’s evaluations and found that the intra- and inter-examiner reliability was excellent. Interestingly, the FES-I, as assessed by the physical therapist, correlated more strongly with the SIAS-motor and FIM total scores than with the patient’s assessment, indicating the FES-I assessed by a physical therapist might better reflect actual movement ability.

A study examining changes in fall-related self-efficacy over time in hospitalized patients with stroke reported improvements in fall-related self-efficacy as their abilities improved (Hellström et al, 2003). On the other hand, it is not clear how fall-related self-efficacy changes in patients with stroke who experience a fall during hospitalization. This study clearly indicates that, regardless of their history of falls, patients with stroke have improved fall-related self-efficacy at discharge compared to that on admission, which may be because the fallers also showed improvement in physical function and ADL during the course of hospitalization. In contrast to our findings, a study that examined changes in fall-related self-efficacy over time in community-dwelling elderly people reported a gradual decline regardless of the history of falls, with a greater decline seen in those who had fallen multiple times (Delbaere et al, 2010). These findings indicate that the perception of falls changes positively with the recovery of physical function in patients with stroke regardless of falls, and the impact of falls may not be consistent with findings in older adults.

The present study suggested that patients with falls overestimated their fall self-efficacy. In addition, the discrepancy in fall risk perception in the single-fall group decreased at the time of discharge from the hospital compared to that at the time of admission, but no change was observed in the multiple-fall group. Reportedly, some falls in hospitalized stroke patients are caused by movements beyond the permitted range based on the patient’s own judgment (Hanger et al, 2014). In a previous study (Walsh et al, 2019) that interviewed people who had repeated falls in the first year after stroke, patients made statements such as prioritizing activities despite being at a risk for falls. Similarly, in another study (Aihara et al, 2021), patients with stroke who experienced a fall during hospitalization were interviewed within a week after the fall; overconfidence in ability, lack of awareness, and lack of anticipation of the fall were cited as causes of the fall, suggesting that stroke patients do not correctly judge their fall risk. In the present study, there was no change in the discrepancy in the perception of fall risk between admission and discharge in the multiple-fall group, suggesting that their inappropriate perception of their fall risk contributed to the recurrence of falls.

The study was conducted in the setting where fall prevention measures were implemented. Therefore, the findings obtained in the present study were influenced by the implementation of counter measures for fall prevention. Especially, education on fall prevention might affect to decrease the discrepancy in patients’ perceptions of in fall risk. It is very interesting to note that, despite this situation, the discrepancy in their perceptions of fall risk remained in the multiple-fall group. In other words, the results of this study pointed that the fall prevention provided in the setting of the present study was insufficient or ineffective for patients who have a large discrepancy in their perceptions of fall risk. Future studies are required to determine whether more rigorous strategies, such as repeated one-on-one therapist education (Haines et al, 2011; Hill et al, 2015), would modify the discrepancy in fall risk perception in patients at high fall risk.

A limitation of this study is that only the history of falls during hospitalization was considered in the present study, as it was difficult to accurately obtain the history of falls prior to hospitalization. Also, its retrospective study design in a single hospital. The generalizability of patients in other institutions cannot be warranted. However, the fall rate of 0.47 per patient (200 falls in 426 patients) in the present study was comparable to the fall rates ranging from 0.22 to 0.95, as reported in previous studies (Nyberg and Gustafson, 1995; Teasell et al, 2002; Czernuszenko and Członkowska, 2009). Furthermore, the characteristics of the fallers in the present study, that is, more severe paralysis and lower ADL, were similar to those observed in previous studies (Nyberg and Gustafson, 1995; Teasell et al, 2002; Czernuszenko and Członkowska, 2009). Considering that the stroke patients included in the present study had similar characteristics to those in previous studies, the findings of the present study may have certain generalizability. A large multicenter study will confirm the findings of this study.

This study revealed that the FES-I assessed by stroke patients differed from that assessed by physical therapists and that the extent of the difference was related to the incidence of falls. The study also revealed that the FES-I did not worsen with the experience of falling but improved as the ability improved. Furthermore, our results suggested that those who fall multiple times may have overestimated their abilities more than the objective viewpoint of a physical therapist.