Systematic review of candidate prognostic factors for falling in older adults identified from motion analysis of challenging walking tasks

Overview of the selected articlesArticle extraction

An overview of the systematic article extraction is given in Fig. 1. In total, 2269 articles were extracted from the three databases. First, 376 duplicates were removed. Another 1790 articles were removed based on their titles and abstracts. After reading the full text, further 25 articles were removed for the following reasons: 23 articles focused on dual-tasks, static balance, single steps or steady walking, one article was a systematic review, and one article lacked the description of the participants. In the end, 78 articles were included in this review.

Fig. 1figure 1

Overview of the systematic article extraction

Fall risk evaluation

Sixteen studies compared a group of higher-risk older adults (mean age ranging from 62.5 to 81.6 years) to a group of lower-risk older adults (mean age ranging from 65.6 to 80.8 years). These studies will be referred to as risk studies. The details of the study designs and populations of risk studies are reported in Table 1. Two older adult performance studies assessed fall risk prospectively [1, 25], by following subjects for one year to determine whether they fall, after they performed the challenging walking task. Six studies assessed fall risk retrospectively [2, 10, 20, 24, 58, 62, 80], by asking subjects at the time of the walking measurement whether they had fallen in the previous months. Three studies evaluated risk based on physical or mental performance at the time of the walking measurement using clinical tests or questionnaires [13, 59, 84]. Two studies evaluated risk based on whether the subjects experienced functional decline or improvement over a one-year follow-up [10, 57]. One study divided the subjects into higher and lower risk depending on whether they fell during the challenging walking task itself [63]. One study compared patients with complaints of “unsteadiness” during walking (higher risk) to a group of healthy controls without a history of falls [15]. The final study compared a group of hospitalised subjects (higher risk) with a group of healthy subjects [11]. Within each study, the two groups were typically age matched, except for two studies [11, 84]. In those two studies, the older adults at higher risk were significantly older than those at lower risk, and this was considered a serious risk of confounding bias (Additional file 3: Appendix C, Table C.1).

Table 1 Description of the risk studies, ordered by gait task

Sixty-two of the selected studies evaluated fall risk based solely on age, comparing a group of younger (mean age ranging from 20.9 to 29.3 years) and a group of older participants (mean age ranging from 55.6 to 81 years). These studies will be referred to as ageing studies in the rest of the text. The details of the study designs and populations of ageing studies are reported in Additional file 4: Appendix D. The full list of references for ageing studies is in Additional file 5: Appendix E.

Tasks

Task characteristics are also reported in Table 1 for risk studies and Additional file 4: Appendix D for ageing studies.

Stair climbing was assessed in 7 risk studies and 19 ageing studies. The stairs had a variety of configurations, ranging from a single [7, 16] to 13 steps [11] and the studies evaluated either ascent, descent or both.

Perturbed walking was assessed in 2 risk studies and 20 ageing studies. Most perturbations were applied through a translation of the support surface (14 studies), either in the mediolateral or anteroposterior direction, or both. The other types of perturbations were waist-pulls [43, 67], ankle pull [9, 50], tripping [63], visual perturbations [40, 75], soapy water [44] and surface drop [37].

Obstacle crossing was studied in 7 risk and 23 ageing studies. The obstacles used had a variety of dimensions, with a height ranging from 0 cm (visually projected obstacles in [12] and [14] up to 30 cm [25, 62] or 30% of the leg length [31, 45, 57, 59, 60]. In the prospective study by Hansson et al., the participants navigated an obstacle course comprising several tasks in sequence: standing up from a chair, walking along a narrow path of 25 cm width for 3 m, walking over an uneven surface, crossing 3 obstacles of 30 cm height, and climbing up and down a stair of 10 steps (Hansson 2021).

In 62 out of 78 studies, subjects were allowed to walk at their preferred velocity.

Most studies used either an infrared camera system (33 studies), force platforms (10 studies), or both (28 studies) to measure walking parameters. One study used an inertial measurement unit (IMU) attached to the right thigh to measure kinematic parameters [25].

Sample size

The selected studies had a wide range of sample sizes, with 30 studies having 10 or fewer participants in one of the groups, and 2 studies having more than 370 participants in total [57, 58]. Overall, risk studies assessed 680 older participants with a higher risk and 927 with a lower risk. Ageing studies assessed 921 younger participants and 1065 older participants.

Level of evidence

An overview of the risk of bias for each domain and article is given in Additional file 3: Appendix C. Based on the seven risk of bias domains, we classified 23 studies with low risk of bias, 48 studies with moderate risk of bias and 7 studies with serious risk of bias. Typical biases among the articles included in the seven domains were:

1.

Confounding factors: confounding due to a difference in the gender ratio between groups (9 studies); no information on the gender ratio between groups (18 studies).

2.

Participant selection: the health status of the participants (physical health, mental health or fall history) was used an exclusion criterion (15 studies); the study population had an unbalanced gender ratio (32 studies, including 11 which included either only females or only males).

3.

Group allocation: risk was evaluated based on balance tests or clinical tests or questionnaires rather than prospective or retrospective fall history (6 studies)

4.

Intended experiment: no information on either randomization or fatigue prevention (42 studies)

5.

Outcome parameters: invalid assessment of centre of mass location (3 studies)

6.

Missing data: missing data due to differences or errors in task performance resulting in unbalanced groups for analysis (10 studies).

7.

Result reporting: significance level was not reported (4 studies).

Motion analysis outcome parameters

A wide range of parameters were reported by the studies, including outcomes related to success, timing, foot and step, centre of mass, force plates, dynamic stability, joints and segments. Table 2 presents the subset of outcome parameters that were reported for at least one risk study, and indicates the studies reporting either significant or non-significant findings for each outcome. Outcome parameters that were only reported in ageing studies are listed in Additional file 6: Appendix F. The most commonly reported parameters were step length (7 risk and 31 ageing studies), stance, swing and compensatory duration (6 risk and 23 ageing studies) and walking, approaching or crossing speed (11 risk and 20 ageing studies). Thirty-seven parameters were reported by a single study in a single task.

Table 2 Significant and non-significant findings in risk studiesLack of agreement between ageing and risk studies

An important finding is that outcomes which were significantly different between younger and older adults were not necessarily good prognostic factors for fall risk (Table 2). This finding was particularly robust when considering outcomes reported by a large number of studies.

The most reported finding was step length, which was found to be significantly shorter between younger and older adults in a majority of studies (Table 2) for stairs (3 out of 4 studies), perturbations (9 out of 12 studies) and obstacles (10 out of 15 studies). In contrast, step length was not significantly different between older adults at higher and lower risk in either stairs [84], perturbations [20, 63] or obstacles [15, 20, 24, 59]. This finding was consistent whether risk was evaluated based on retrospective fall history [20, 24, 62], balance performance [63], physical and mental level [59, 84] or unsteadiness complaints [15].

Another commonly reported finding was speed when walking over obstacles (Fig. 2). The speed was found to be significantly higher in younger than older subjects in 8 out of the 10 studies which assessed this parameter (Table 2). In contrast, no significant difference in obstacle walking speed between older adults at higher and lower risk was found in any of the 5 studies that assessed this parameter, whether risk was evaluated based on retrospective fall history [24, 62], physical level [59], unsteadiness complaints [15] or walking decline over a year [10].

Since the goal of this review was to identify candidate fall risk prognostic factors for older subjects, in the rest of this result section we only report the results from risk studies.

Walking, approaching, and crossing speed

In most studies (62 out of 78), subjects were allowed to walk at self-selected speed. Walking, approaching or crossing speed was assessed in 5 risk studies with stairs, 1 risk study with perturbations and 5 risk studies with obstacles. Stair studies reported speed in number of steps per second [1, 2, 10] and two studies reported the total stair ascent or descent time [57, 58]. The perturbation study reported speed in meters per second [63]. Three obstacle studies used multiple steps (including the obstacle crossing steps) to assess an average walking speed, either over several meters or over 6 steps [10, 24, 62]. One obstacle study reported the crossing speed of the single stride over the obstacle [59].

As mentioned previously, none of the obstacle studies found a significant difference in speed between older adults at higher and lower risk (Fig. 2). While subjects significantly slowed down when crossing obstacles with increasing height, this did not differ significantly between fall risk groups [59]. One study reported the decrease in walking speed when walking over an obstacle compared to normal walking [10]. Whereas the walking speed over the obstacle itself was not significantly different across groups, the decrease in speed compared to baseline was significantly larger for older subjects whose walking speed declined over a one-year follow-up [10]. Note that in this study, retrospective fall history was also assessed, and was not significantly different between subjects whose walking speed improved or declined over one year.

Fig. 2figure 2

Forest plot of walking speed. Circles: open blue– younger adults, yellow closed– older adults with low fall risk, red closed – older adults with high fall risk. For Brach (2011): older adults whose gait speed improved / stayed the same / deteriorated after a year are indicated in green / yellow / red.; m/s, meters per second; N.A.*, not applicable, the standard deviation was not given; n.s., not significant. Ntot is the total number of participants

Fig. 3figure 3

Foot clearance definitions from different studies. Some studies reported overall foot clearance, while others reported heel clearance and/or toe clearance. These clearances were reported as a vertical distance or as a horizontal distance between participant and stair or obstacle while the foot is in the air. Furthermore, clearance of the leading leg and the trailing leg were reported. V, vertical; H, horizontal

Fig. 4figure 4

Forest plot of heel/foot/toe clearance. Circles: open blue– younger adults, yellow closed– older adults with low fall risk or non-fallers, red closed – older adults with high fall risk or fallers. Ntot is the total number of participants. n.s. is non-significant finding

Walking speed on stairs was not found to be significantly different between older adults at higher and lower risk when fall risk was assessed prospectively [

留言 (0)

沒有登入
gif