Footwear and Falls in Long-Term Residential Aged Care Facilities: An Analysis of Video Capture Data

Falls in older people are a major public health problem [1] and are responsible for the majority of wrist [2] and hip fractures in this population [3]. In long-term residential aged care, approximately one in two residents falls more than once every year [4], which is up to 3 times higher than the rate in community-dwelling older people [5]. Although several risk factors for falls in long-term care have been identified [6], they remain a major cause of morbidity and mortality [7], and the effectiveness of multifactorial interventions to prevent falls remains uncertain [8].

Falls result from a complex interaction between the individual and their environment [1]. One potentially modifiable risk factor that may contribute to falls in long-term care is footwear. The most frequently worn indoor footwear are slippers, which often comprise design features considered detrimental to balance, including lack of fixation, thick and soft midsoles, and smooth outersoles [9‒12]. It has been shown that wearing slippers is a risk factor for falls [13‒15] and fall-related injury [16‒19]. Going barefoot or wearing socks is also common and has been associated with cautious gait patterns [20, 21] and falls [19, 22, 23] compared to being shod. Consequently, older people are often advised to wear shoes indoors where possible [24], although it has been acknowledged that the evidence for recommending footwear changes in those at risk of falls is not strong, as few randomised trials have been undertaken [25]. The most recent guidelines specify that footwear should be assessed but do not recommend one type of shoe over another [26].

Much of our understanding of falls is based on incident reports or interviews [27‒29], and as many falls are not observed [30, 31], the circumstances responsible are uncertain [32] and few are biomechanically analysed in real world scenarios. This is particularly true for establishing a link between footwear and falls, as footwear is a situational risk factor, in that it is likely to only play a role at the time of the fall [33]. Video technology provides a means of capturing real-life falls [34] and has been used to ascertain factors involved in stepping [35], transfer-related falls [36], and fall injury [37]. In this study, we used a database of video recordings of falls in long-term care to determine whether footwear contributed to the fall. In doing so, we hoped to clarify the role of footwear and to provide recommendations as to what type of footwear older people should wear in this setting.

This study used an existing dataset of video recordings that have been made available to other researchers through a data sharing platform (Databrary) hosted by New York University [38]. Approvals to access this dataset were provided in an ethical application to the Research Office at La Trobe University, Melbourne, Australia (HREC 22229). The original study was conducted between 2007 and 2010 by Dr. Stephen Robinovitch under full ethical approval from his institution, the Office of Research Ethics at Simon Fraser University, British Columbia, Canada [34].

Participant characteristics were documented according to age, sex, activities of daily living self-performance scale (including four items: eating, locomotion, toileting, and personal hygiene, with scores ranging from 0 [independent] to 6 [total dependence]) [39], the cognitive performance scale (including 5 items: short-term memory, daily decision making, ability to make self-understood, eating self-performance, and comatose status, with scores ranging from 0 [intact] to 6 [very severe impairment]) [40], vision, with scores ranging from 0 (adequate) to 4 (severely impaired), and number of medications per week.

Falls were defined as “an unexpected event in which the resident comes to rest on the ground, floor, or lower level” [41]. All individuals captured falling were residing in one of two long-term residential aged care facilities (a 312 bed facility in Delta and a 236 bed facility in Burnaby) in British Columbia, Canada, where the falls occurred [34]. In Delta, 45% of falls occurred in common areas, of which 65% were captured on video. In Burnaby, 34% of documented falls occurred in common areas, of which 28% were captured on video. All captured falls occurred in common areas (lounges, dining rooms, and hallways). Each individual who was recorded during falling, or their decision maker, provided informed consent for sharing their fall video footage. Permission to share both videos and medical records was provided by 98 participants, while consent for sharing of their videos only was provided by 20 participants (resulting in a total sample of 118 participants). In cases where a fall was captured by more than one camera, the views from the individual cameras were combined into a single video file. Most falls (n = 158) were captured from only one camera view. For 99 falls, videos were included from two camera views, and for 43 falls, videos were included from three or more camera views. The falls were recorded on video at a frame rate between three and 33 frames per second. The camera resolution ranged from 352 × 240 pixels to 1,920 × 1,080 pixels.

We used data related to surface (linoleum/vinyl tiles, or carpet), cause (incorrect shift of bodyweight, loss of support with an external object, trip/stumble, falling asleep/legs collapsing/loss of consciousness, slip, or hit/bump), activity at the time (walking, standing, seated/wheeling in wheelchair, transferring from a sitting or lying position or transferring from a standing position), use of mobility aid (none, wheelchair in use, walker in use, wheelchair visible but not being used, none visible belonging to the individual, walker visible but not being used, walker visible, and cane in use), and fall direction (forwards, backwards, sideways, straight down) [42].

The primary author (H.B.M.) screened all 300 falls and identified those where the footwear could be identified, and documented these as slippers/sandals, shoes, socks or barefoot. Following this, the number of falls where footwear potentially contributed (defined by the question “Could the fall have potentially been averted if different footwear was worn?”) was determined, and the potential reason for the fall was documented as inadequate slip resistance (the sole is too slippery, resulting in a backwards, forwards or sideways slip); excessive slip resistance (the sole grips the ground and arrests forward momentum); inadequate fixation (the foot is not held into the shoe due to inadequate laces, buckles, Velcro™, etc.); mediolateral instability (the shoe is too soft/unsupportive and collapses in a medial or lateral direction); inadequate toe clearance (absence of anterior rocker results in toe contacting ground during swing, leading to a trip); and instability due to heel elevation (high heels result in anteroposterior or mediolateral instability).

Three other researchers (S.M.B., a podiatrist, J.A.M., a physiotherapist and S.E.M., a podiatrist) were then asked to independently view the relevant videos and state whether they agreed with the assessment and document their views on the reason for the fall. A consensus meeting was then held, with a predetermined agreement threshold of 75% [43].

Falls rather than fallers were considered the unit of analysis. Statistical analysis was undertaken using IBM SPSS Statistics version 26.0 (IBM, Armonk, NY, USA), and involved reporting of descriptive data and frequencies. Comparison of means between those where footwear potentially contributed to the fall and those where footwear did contribute to the fall was undertaken using the independent samples t-test, and differences in frequencies between these groups was undertaken using cross-tabulations and the χ² statistic. The cut-off for significant differences was p < 0.05. The Sankey plot was developed using SankeyMATIC (https://sankeymatic.com).

The dataset comprised 300 falls experienced by 118 older adults (57 women accounted for 175 falls, and 61 men accounted for 125 falls). At the time of falling, the individuals ranged in age from 58 years to 98 years (mean age 82.8 years, SD 7.6). Table 1 reports the characteristics of fallers according to the role of footwear. Falls where footwear was considered to contribute involved more females (75 vs. 56%; χ² = 5.3, p = 0.022) and occurred in older people with better activities of daily living score (t = −2.2, p = 0.028) than falls unrelated to footwear.

Following the initial screening of 300 falls, 76 (25%) were excluded as it was not possible to accurately ascertain the footwear worn at the time, due to either the foot being obscured or the video quality being insufficient. Of the remaining 224 falls, the most commonly worn footwear was slippers/sandals (100; 45%) or shoes (99; 44%), followed by socks (19; 8%) and being barefoot (6; 3%).

Of the 224 falls where the footwear worn at the time could be ascertained, 44 (20%) were deemed to be potentially related to footwear after initial screening. These 44 falls were then screened independently by three additional assessors, where there was complete (100%) agreement for 16 (36%) falls, 75% agreement for 23 (52%) falls, and 50% agreement for 5 (11%) falls. Following the consensus meeting, there was complete (100%) agreement for 31 (70%) falls, 75% agreement for 9 (20%) falls, and 50% agreement for 4 (10%) falls. Therefore, of the original 44 falls that were identified as potentially related to footwear, 40 (89%) reached the predetermined threshold agreement of 75%. The remaining 4 falls were subsequently considered to be unrelated to footwear, so the final proportion of falls potentially related to footwear was 18% (40 out of 224).

Of the 224 falls where footwear worn could be ascertained, 210 (94%) took place on linoleum/vinyl tiles and 14 (6%) took place on carpet. The direction of falls was primarily backward (91; 41%), followed by primarily sideways (61; 27%), primarily forward (41; 18%) and straight down (31; 14%). The cause was deemed to involve incorrect shift of bodyweight (122; 54%), loss of support with an external object (42; 19%), a trip/stumble (36; 16%), 16 (7%) falling asleep/legs collapsing/loss of consciousness (16; 7%), a slip (6; 3%) and a hit/bump (2; 1%). Activity at the time involved walking (86; 38%), standing (47; 21%), seated/wheeling in a wheelchair (32; 14%), transferring from a sitting or lying position (30; 13%) and transferring from a standing position (29; 13%). No mobility aid was present in 96 falls (43%), followed by wheelchair in use (49; 22%), walker in use (31; 14%), wheelchair visible but not being used (18; 8%), none visible belonging to the individual (14; 6%) walker visible but not in use (9; 4%), and walker visible suspected to belong to the individual but not being used (2; 1%).

Characteristics of falls according to role of footwear are shown in Table 2. The 40 falls that reached agreement for being potentially related to footwear were significantly more likely to involve socks (χ² = 44.1, p < 0.001) and a walker in use (χ² = 5.1, p = 0.024), and less likely to involve shoes (χ² = 14.1, p < 0.001), falling straight down (χ² = 7.5, p = 0.006), and involve a wheelchair in use (χ² = 4.0, p = 0.045).

Of the 40 falls that reached agreement for being potentially related to footwear, 17 (43%) involved slippers/sandals, 14 (35%) involved socks, 7 (18%) involved shoes, and 2 (5%) involved being barefoot. Across all 224 falls where footwear could be ascertained, the likelihood of footwear contributing to the fall was highest for socks (14 out of 19 falls; 74%), followed by being barefoot (2 out of 6 falls; 33%), slippers/sandals (17 out of 100 falls; 17%), and shoes (7 out of 99 falls; 7%) (see Figure 1).

Figure 2 shows a Sankey plot which demonstrates the complex relationship between footwear type, fall reason and fall direction. As can be seen from the diagram, slippers/sandals (the most common reason for falls in the 40 footwear-related falls, responsible for 17 falls) most frequently resulted in mediolateral instability (8; 47%), followed by inadequate fixation (4; 23%), inadequate slip resistance (3; 18%) and inadequate toe clearance (2; 12%). Socks, the next most common reason (14 falls), most frequently resulted in inadequate slip resistance (13; 93%) followed by excessive slip resistance (1; 7%). Shoes, responsible for 7 falls, resulted from inadequate slip resistance (3; 43%) followed by inadequate toe clearance (3; 43%) and (1; 14%). Finally, barefoot, responsible for 2 falls, was solely due to inadequate slip resistance (2; 100%).

As can be seen from the plot, fall direction was evenly split between sideways (14; 35%), backward (14; 35%) and forward (12; 30%) falls. Sideways falls resulted from mediolateral instability (7; 50%), inadequate slip resistance (6; 43%) and inadequate fixation (1; 7%), forward falls resulted from inadequate toe clearance (5; 42%), inadequate slip resistance (3; 25%), inadequate fixation (2; 17%), mediolateral instability (1; 8%) and excessive slip resistance (1; 8%), while backwards falls primarily resulted from inadequate slip resistance (12; 86%) followed by inadequate fixation (1; 7%) and mediolateral instability (1; 7%).

The objective of this study was to use a database of video recordings of falls in long-term care to determine whether footwear contributes to falls. From 224 falls, we found that the most common footwear worn at the time of the fall was slippers/sandals or shoes, but that the likelihood of footwear contributing to the fall was highest for socks. The “typical” fall in this population occurred in women on linoleum/vinyl tiles, involved incorrect shift of bodyweight while walking without a mobility aid and primarily involved falling backwards.

Before discussing the findings in detail, it is worth noting that footwear could be adequately ascertained in only 75% of falls, due to either poor video quality (e.g., distance between the faller and the camera) or the footwear being obscured by the older person or furniture. Ascertainment of footwear from video recordings has previously been reported as potentially problematic, as out of 24 observations of biomechanical, situational, behavioural, and environmental aspects of falls, footwear type had relatively low inter- and intra-rater reliability [42]. In the current study, we excluded 25% of falls where footwear could not be confidently determined, so we are uncertain as to whether these falls were related to footwear.

The most common cause of footwear-related falls was inadequate slip resistance, followed by mediolateral instability, inadequate toe clearance, inadequate fixation, and excessive slip resistance (see Fig. 2). Perhaps not surprisingly, no falls were related to instability due to heel elevation, as no participants wore high heels, and it has previously been shown that older women wear lower heels as they age [44]. The relatively high rate of falls related to inadequate slip resistance of footwear may appear to be inconsistent with the relatively small number of falls attributed to slipping overall; however, the most common cause of falls was incorrect shift of bodyweight, and most falls occurred while walking. Taken together, these findings suggest that inadequate slip resistance of footwear may be a largely hidden but potentially modifiable risk factor in this population.

Sock-related falls most frequently resulted from inadequate slip resistance, and most commonly resulted in falling backwards. Socks have been shown to limit sensory feedback from the plantar surface of the foot [45] and as a consequence, gait patterns are more cautious in socks than barefoot [46]. There is a higher rate of falls in those who wear socks compared to shoes [19, 22, 23], and a review of emergency department records suggests that the risk of head injury is markedly increased in older people who fall backwards [47]. As a result of these observations, many hospitals and aged care facilities have introduced “non-slip” rubber or synthetic tread socks to improve slip resistance. However, a recent review found no strong evidence that non-slip socks are beneficial [48], so it may therefore be advisable for older people in long-term care to avoid wearing socks in general.

The lowest rate of footwear-related falls occurred in those wearing shoes. Although we were unable to document the features of individual shoes, our findings are consistent with previous studies in community-dwelling older people [22, 23] and support the recommendation that older people at risk of falls should wear shoes indoors [24]. Enclosed shoes generally provide better slip resistance than slippers, socks, or going barefoot, and they may be protective against future falls. We acknowledge, however, that this recommendation may be difficult to implement, as many older people base their footwear purchases on comfort [9] and wear slippers indoors [9‒12]. As a pragmatic solution, it may be possible for older people who are not prepared to wear shoes indoors to instead wear slippers designed to maintain balance, as backless slippers impair balance [49] and provide a greater risk of fall-related injury [15] than enclosed slippers.

The role of mobility aids in footwear-related falls has not been thoroughly investigated. We found that falls potentially related to footwear were less likely to involve a wheelchair but more likely to involve a walker. The negative association between footwear-related falls and wheelchair use is not surprising, given that footwear is likely to play a role when balance is challenged when performing standing tasks. However, the interaction between footwear-related falls and the use of walkers is both novel and interesting. Walkers, although generally considered beneficial for balance, may also increase attentional, physiologic, and metabolic demands [50], thereby increasing the risk of fall-related injury in some individuals [51]. This may be particularly evident in footwear-related falls.

The findings of this study need to be considered in the context of several inherent limitations. First, although the use of video technology to document falls is an improvement on previous incident report methods, we acknowledge that falls were only analysed in common areas such as dining rooms, lounges, and hallways, and a relatively large proportion of falls were not captured. Furthermore, in one-quarter of falls we were unable to ascertain the footwear worn at the time. Second, the primary author screened all 300 falls before presenting 44 of these for consensus. This was a pragmatic decision due to time constraints but is partly ameliorated by the fact that where any doubt existed, footwear was considered to not be involved in the fall. The results are therefore likely to be a conservative estimate of the number of falls related to footwear. Third, we documented footwear into 4 categories (slippers/sandals, shoes, socks or barefoot), which was feasible but too rudimentary to delineate individual features of shoes. Finally, falls were analysed using an existing database obtained from two long-term residential aged care facilities where the prevalence of physical and cognitive impairment was quite high. As has been previously recognised, findings from this study may not be transferable to healthier populations of older adults, or for falls in different settings [34].

Despite these limitations, our findings suggest that footwear does potentially contribute to a substantial number of falls in long-term care. Ideally, older people should be advised to wear enclosed shoes indoors, but where this is not possible or desirable, indoor footwear with safety features such as firm slip resistant soles, and an enclosed upper may be acceptable. Wearing socks, however, would appear to place an older person at risk of future falls and should therefore be a target for intervention in this population.

The authors would like to thank Dr. Stephen Robinovitch for making the data available.

This study used an existing dataset of video recordings that have been made available to other researchers through a data sharing platform (Databrary) hosted by New York University. Approvals to access this dataset were provided in an ethical application to the Research Office at La Trobe University, Melbourne, Australia (HREC 22229). The original study was conducted by Dr Stephen Robinovitch under full ethical approval from the Office of Research Ethics at Simon Fraser University, Burnaby, British Columbia. Each resident or proxy provided written permission to the facility to acquire video footage in common areas.

The authors have no conflicts of interest to declare.

H.B.M. was a National Health and Medical Research Council Senior Research Fellow (ID: 1135995) at the time this study was undertaken. The funder had no role in the design, data collection, data analysis, and reporting of this study.

H.B.M: study conception, acquisition of dataset, data collection, analysis, interpretation, and drafting of manuscript. S.M.B., J.A.M., and S.E.M.: data collection and interpretation. All authors read and approved the final manuscript.

This study used an existing dataset of video recordings that are available to other researchers through a data sharing platform (Databrary) hosted by New York University.