Early Heelstrike Kinetics Are Indicative of Slip Potential During Walking Over a Contaminated Surface

Abstract

Objective: The objective of this study is to examine ground kinetics early in stance while walking on a contaminated surface and assess the potential of kinetics to quantify risk of slipping.

Background: Prior studies of slipping have dismissed early ground kinetic data, and therefore no prior literature has been able to assess the viability of using these data to quantify slip potential.

Method: A total of 11 healthy male participants volunteered to walk over a force plate that was at random times contaminated with soap. Ground kinetics were measured by the force plate (2400 Hz), and heel displacement was quantified using high-speed video cameras (240 Hz) and retro-reflective markers.

Results: The results indicated a significant reduction in shear force as early as 0.42 ms after heelstrike for contaminated trials, whereas for utilized coefficient of friction, a significant reduction was not seen until 11.34 ms. Heel displacements considered “safe” in the literature (< 30 mm) demonstrated proportionally different thresholds for shear force and utilized coefficient of friction.

Conclusion: The authors suggest that shear force in early stance shows more promise in quantifying slip potential as compared to utilized coefficient of friction given that (a) significant differences are seen earlier in shear than utilized coefficient of friction and (b) the threshold for utilized coefficient of friction, over which heel displacement stabilized to a “safe” value, exceeded values for utilized coefficient of friction that have been recommended as “safe.”

Application: These results have wide implications for standards related to the design and testing of interventions to prevent injuries because of slipping.

Keywords

slips and falls injury prevention friction measurement

Introduction

Slip and fall injuries constitute a public health hazard that is costly to society. It has been stated that approximately 25% of annual workplace slip-, trip-, and fall-related injuries result in more than 31 lost workdays at a cost of $10 billion per year to the U.S. economy (Yoon & Lockhart, 2006). Clearly, any intervention that can mitigate the cost of slips would provide a substantial savings to health care and society at large. However, as stated by Grönqvist, Chang, et al. (2001), slip and fall events are dependent on many factors, and any intervention must first seek to understand how these aspects can create a slip hazard.

One factor that is considered critically important is the underfoot condition, or surface, on which the person is walking (Grönqvist, Chang, et al., 2001). In particular, the frictional interaction between the shoe and the floor surface has been widely studied in the context of slipping, and the coefficient of friction (CoF) between the shoe and surface, defined as the ratio of shear force to normal force, is often presented as a variable of interest. When CoF is measured during gait, it is referred to as utilized coefficient of friction (UCoF). Attempts have been made to correlate the severity of slips with changes in the UCoF during gait (Grönqvist, Hirvonen, & Matz, 2001; Grönqvist, Hirvonen, & Tuusa, 1993; Hanson, Redfern, & Mazumdar, 1999; Perkins, 1978; Powers et al., 2007; Redfern & Rhoades, 1996; Strandberg, 1983, 1985; Strandberg & Lanshammar, 1981); however, it is still unclear from these studies how to consistently identify a slip hazard based on a standardized measure of UCoF.

One key limitation in the study of UCoF has been consistency in the time point during gait at which UCoF is measured. Researchers have examined a variety of timings ranging from absolute time points of 50 ms up to 200 ms following heelstrike, or time-averaged values over one, or many steps (Cham & Redfern, 2002; Grönqvist et al., 1993; Hanson et al., 1999; McVay & Redfern, 1994; Perkins, 1978; Strandberg, 1983, 1985; Strandberg & Lanshammar, 1981). More recently, McGorry, DiDomenico, and Chang (2010) concluded that slip and nonslip trials may be differentiated based on foot displacement and UCoF values at 25 ms following heelstrike. From the literature reviewed, it seems as though the shear and vertical forces acting on the foot at heelstrike have not been explicitly studied. However, Newtonian mechanics dictate that the movement (or slip) of the foot must be a consequence of these forces acting on it. In other words, the amount of slip occurring at 25 ms after heelstrike must depend on the ground reaction forces (GRFs) and UCoF prior to this 25 ms time point. These early heelstrike vertical and shear force data, however, are not currently available.

Therefore, the purpose of this study was to examine GRFs during walking and determine whether they are reduced immediately following heelstrike with a slippery surface condition. It was hypothesized that GRFs and the CoF immediately following heelstrike would be reduced with a slippery surface condition.

Method

A total of 11 healthy adult males, free of musculoskeletal injury, were recruited for this study (age = 29.5 ± 5.7 years, mass = 75.6 ± 10.4 kg, height = 175.5 ± 7.7 cm). The study received approval from the institutional ethics committee, and all participants gave informed written consent prior to participation.

A scaffold was erected over an in-ground force plate as part of a fall-arrest system that included a harness and lanyard (Figure 1). Each participant wore a pair of work boots (Dakota 529, Mark’s Work Wearhouse, Canada) in an appropriate size. Before data collection began, each participant was strapped into the harness and instructed about its operation. The participants were also asked to fall into the harness so that they would trust its operation and be less concerned about falling.

Figure 1.

Schematic of the experimental setup.

The participant wore a harness that was secured to an overhead trolley. The trolley rolled along an overhead beam supported by a scaffold system. A pulley system was used to ensure that the overhead trolley remained above the participant at all times, which in turn ensured that the lanyard would catch the participant properly if he fell. Participants walked over an in-ground force plate and through a 3-D capture volume while kinematics and kinetics were recorded.

All participants were then asked to perform 3 to 5 practice walking trials across the lab to ensure that they would be contacting the force platform with the correct foot. The participants were given modified safety goggles, the bottom half of which was obstructed. The goggles impeded their view of the ground but allowed them to look ahead and walk in a natural manner. This was done to prevent the participants from being aware of the floor surface condition. The participants were then informed that the test would begin and that at random intervals the force plate would be contaminated with dish soap, making it slippery (termed a “contaminated trial”). Between trials, participants were asked to face away from the force plate and wear ear protectors so they could not see or hear if any change in surface condition occurred. During this time, the force plate was either left untouched or contaminated using dish soap applied with a rag in an even layer approximately 1 to 3 mm thick on the force platform. Following a contaminated trial, the normal dry surface was restored by thoroughly wiping away the soap with a clean, damp rag and then drying with a clean dry rag. A minimum of 4 contaminated trials and 10 dry surface trials were collected for each participant. Pilot testing demonstrated that the use of the goggles was effective in preventing anticipatory reactions as no differences were found between the first contaminated trial and subsequent contaminated trials.

For each trial, three-dimensional GRFs were measured using a force platform at 2400 Hz (Kistler AG, Winterthur, Switzerland) mounted flush with the lab floor. Heelstrike was defined as the point at which the vertical GRF crossed a rising threshold of 3 N. To quantify any slips that occurred, 3-D foot kinematics were collected at 240 Hz using three passive, retro-reflective markers and a motion capture system consisting of eight high-speed digital cameras (Motion Analysis Corporation, Santa Rosa, California). The digital cameras were calibrated before each session with a residual error less than 0.5 mm. The reflective markers were placed on the right boot of each participant in the following locations: two markers on the posterior aspect of the boot upper, one immediately above the outsole and the other approximately 5 cm directly above the first, and one marker on the lateral aspect of the boot upper, immediately above the outsole (Figure 2). Based on the known geometry of the three markers and each work boot, a fourth “virtual marker” was reconstructed at the heel bevel (the most posterior and inferior point of the boot outsole). This “virtual” heel bevel marker was used to quantify horizontal heel displacement during the walking trials.

Figure 2.

Photograph of the boot used in the experiment, showing marker placement. The “virtual marker” at the heel bevel is indicated by the shaded circle.

Kinematic data were processed using Expert Vision Real Time software (Motion Analysis Corporation, Santa Rosa, California) to determine the position of each marker in 3-D space. Three-dimensional marker coordinates and digital signals from the force platform were then imported into MatLab (The Mathworks, Natick, Massachusetts) for further analysis. A low pass wavelet filter was used to filter kinematic data (25 Hz cutoff) and kinetic data (50 Hz cutoff). Heel displacements were calculated as the resultant difference in the heel bevel marker position from heelstrike to 205.8 ms following heelstrike. This time point was chosen as it was late enough to measure larger slips but occurred prior to the push-off phase. After smoothing, GRFs for the first 25 ms of ground contact were extracted for further processing as this time period comprised the early stance kinetics that were the focus of the current work. Shear force acting on the foot was calculated as the resultant shear in the plane of the ground, whereas normal force was taken as the vertical component of the GRFs. UCoF was calculated by dividing the shear force by the normal force at each time point. Both shear force and heel displacement were calculated as absolute resultant quantities, under the assumption that the two quantities would always occur in opposition, and therefore direction information was not strictly necessary.

Statistical analyses were conducted between contaminated and dry surface trials for average values of each variable at particular time points. For heel displacement, averages for each participant were calculated at 205.8 ms postheelstrike. For kinetics, averages for each participant were calculated at 0.42 ms intervals from heelstrike to 25.2 ms postheelstrike. These data were then compared across surface conditions using paired-data methods. During the initial analyses, it was observed that the data exhibited nonnormal distributions of the paired-differences between contamination conditions. It was therefore deemed appropriate to use a nonparametric statistic, the Wilcoxon signed-rank test, at every time point, for each variable independently. Holm’s procedure was then used to adjust each p value prior to comparison with the critical p value (p = .05). All statistical analyses were completed in MatLab.

Results

Heel displacement was significantly greater for contaminated trials at 205.8 ms after heelstrike (Figure 3). For 49 contaminated trials, 27 trials resulted in heel displacements larger than 10 mm, whereas no trials resulted in falls. Table 1 shows a breakdown of slip distances based on categorizations used by Perkins (1978) and Cham and Redfern (2002). These categorizations from the literature were used to assess foot movement in terms of “slip” to establish a threshold for potentially hazardous foot movement. The five trials that resulted in distances larger than 100 mm did not actually result in falls.

Figure 3.

Graph comparing average heel displacement at 205.8 ms following heelstrike between dry and contaminated trials. *Significant difference (p < .05) between surface conditions.

Table 1:

Breakdown of Heel Displacements for Contaminated Trials, by Distance Traveled Between Heelstrike and 205 ms Following Heelstrike

	Micro Slip (< 10 mm)	Macro Slip (10–100 mm)	“Fall” (> 100 mm)
# of trials	22	22	5

Note. Categorizations are based on Perkins (1978) and Cham and Redfern (2002).

Sample GRFs and heel displacement data for one participant are shown in Figure 4. In general, during the 205.8 ms following heelstrike, contaminated trials showed decreased normal and shear forces, decreased UCoF values, and increased heel displacements. GRF peaks occurred later in stance, with smaller magnitudes for the slippery surface condition.

Figure 4.

Sample data from one participant demonstrating general differences in kinetics and kinematics between dry and contaminated trials. Traces are smoothed data from one trial in a dry surface condition (solid) and one trial in a contaminated condition (dashed). F_vertical is the vertical ground reaction force, whereas F_anterior represents the anterior ground reaction force.

Significant differences were found between dry and contaminated trials for UCoF and shear force, shown in Figure 5; however, no significant differences were found for normal force. The timing of differences in kinetics varied for each quantity. For contaminated trials UCoF was significantly less from 11.34 ms to 21.0 ms after heelstrike as compared to the dry surface trials. Shear force was significantly lower in the contaminated trials from 0.42 ms to 21.84 ms after heelstrike.

Figure 5.

(Top) The time history of utilized coefficient of friction (UCoF) for the first 25.2 ms of stance for dry (solid) and contaminated (dashed) trials. (Bottom) The time history of shear force for the first 25.2 ms of stance for dry (solid) and contaminated (dashed) trials. For both panels the traces are averages across all participants, with error bars indicating one standard deviation for each time point. The asterisks indicate significant differences (p < .05) for time points bound by the brackets.

Discussion

The purpose of this study was to examine GRFs during walking and determine whether they are reduced during the first 25 ms of ground contact with a slippery surface condition. Significantly reduced UCoF values were seen for a slippery surface condition as early as 11.34 ms, indicating that kinetic changes are occurring prior to those time points examined in previous literature. Changes in shear force were seen even earlier, at 0.42 ms after heelstrike. These statistical changes occurred for a slippery surface despite the fact that 45% of heel displacements in this study were less than 10 mm, and no displacements resulted in falls. It therefore seems that early ground kinetics may actually overestimate the slip potential of a particular surface condition and therefore can serve as a conservative indicator of a suboptimal traction situation.

The results for heel displacement and ground kinetics in the current study are in good agreement with existing slip literature. Slips have generally been characterized by large heel displacements (>10 mm; Brady, Pavol, Owings, & Grabiner, 2000; Cham & Redfern, 2001, 2002; DiDomenico, McGorry, & Chang, 2007; Grönqvist, 1999; Leamon & Li, 1990; Perkins, 1978; Strandberg, 1983). Meanwhile, GRFs were reduced for the contaminated condition, a finding presented by other authors (Cham & Redfern, 2002; Strandberg, 1983). For the current study, the normal force was not significantly different, potentially because of the low severity of the slips that occurred. These kinematic and kinetic results are likely interrelated; a reduction in shear force because of the contamination of the floor would reduce the braking impulse that contributes to the deceleration of the foot to zero velocity. This would result in the increased heel displacements seen during midstance.

Nearly all prior studies of slipping chose not to present gait kinetics and kinematics in early stance, a gap in the literature that is addressed in the current study. The reason for the lack of prior data could perhaps be attributed to an early study of slipping authored by Perkins (1978). Using a force platform and stroboscopic photography, Perkins attempted to elucidate the causes of slipping during walking by measuring gait kinematics and kinetics. In this study, however, the author dismissed kinetic data that were acquired shortly after heelstrike, citing excessive noise in the signal. The author suggested that this was a result of equipment limitations and that this portion of the gait cycle should be discarded as it could provide no useful information. Perkins therefore gravitated to a time period of 50 to 100 ms, a strategy that seems to have been subsequently adopted by researchers in this area. In contrast, the current results suggest that modern equipment, paired with a sufficiently large participant pool, allows for the detection of kinetic changes much earlier than previously thought.

A wide range of UCoF values corresponding to slip events has been presented in the literature, making comparisons with the current results difficult. Values have been measured as low as 0.02 for slip falls, whereas some have been as high as 0.15 for slip recoveries (Grönqvist et al., 1993; Strandberg, 1983). Because of the fact that (a) UCoF values are reported at different time points during the gait cycle and (b) UCoF is, by definition, dependent on the surface conditions, it is impossible to draw rigorous conclusions by comparing values across studies. What is clear and consistent is that UCoF decreases when surface friction is decreased (Cham & Redfern, 2002; Chang et al., 2001; Grönqvist et al., 1993; Manning & Jones, 2001; Redfern & DiPasquale, 1997; Strandberg, 1983; Strandberg & Lanshammar, 1981). The current study supports this general conclusion, and in addition, we can observe that the differences in UCoF occurred as early as 11.34 ms after heelstrike. Values of UCoF in early stance must be interpreted with caution, however, as they exhibit very large dispersion, likely because of noise in the base quantities of normal and shear forces. In this respect, shear forces are a superior measurement as they exhibit much less dispersion, making the detection of kinetic changes more reliable.

Comparing the panels in Figure 6, it is clear that although early UCoF and shear force vary with surface condition, shear force may provide a more robust threshold for the determination of slip potential. It is apparent that for shear forces greater than 30 N, displacements stabilize to less than 30 mm, a displacement that is considered safe (Cham & Redfern, 2002; Leamon & Li, 1990; Perkins, 1978). Although displacements stabilize for UCoF as well, the threshold is rather high at approximately 0.95, which is well above values for UCoF that have been previously considered “safe” (i.e., a required CoF between 0.13 and 0.50, found for a wide range of surface conditions; Barnett, 2002; Brungraber, 1976; Cham & Redfern, 2002; Cooper, Prebeau-Menezes, Butcher, & Bertram, 2008; Francis & Zozula, 1990; James, 1980; Perkins, 1978; Redfern et al., 2001; Strandberg, 1983; Strandberg & Lanshammar, 1981). Given previous thinking on what is “safe,” it seems as though our current understanding of UCoF is insufficient to explain this result. Shear force could therefore be superior to UCoF as a predictor of slipping outcomes, as the probability of heel displacements greater than 30 mm (i.e., slips) increased precipitously when shear force was less than 30 N at 11.34 ms following heelstrike. It is interesting that although certain dry trials exceeded the threshold of 30 mm for displacement, they also exhibited shear forces that were below the shear force threshold of 30 N, indicating that the methodology seems to hold for an uncontaminated surface. In addition, since this threshold of 30 N is a force and not a CoF, the threshold is independent of surface-shoe conditions and can be applied to a variety of scenarios, in contrast to a threshold for UCoF, which is dependent on the shoe and surface. Indeed, another recent article on the topic has called into question the validity of CoF measures across various shoe and surface conditions (Kim & Nagata, 2008). The adoption of shear force as a method to study slipping could potentially allow valid interstudy comparisons to be made, a method that seems intractable with UCoF.

Figure 6.

(Top) Scatterplot of heel displacement at 205.8 ms after heelstrike versus utilized coefficient of friction (UCoF) at 11.34 ms after heelstrike. (Bottom) Scatterplot of heel displacement at 205.8 ms after heelstrike versus shear force at 11.34 ms after heelstrike. For both panels, each point in the scatterplot represents one trial, either dry (dark diamond) or contaminated (light square). The horizontal dashed line represents a displacement threshold of 30 mm, considered “safe” (Cham & Redfern, 2002; Leamon & Li, 1990; Perkins, 1978). The vertical dashed lines represent the thresholds for UCoF and shear force above which displacement stabilizes to less than 30 mm.

In conclusion, this study found that shear force and UCoF during walking demonstrated significant decrease as early as 0.42 ms and 11.34 ms after heelstrike with a slippery surface condition. These differences occur at an earlier time point than has been previously reported. Furthermore, although shear force exhibited a sensible threshold with respect to heel displacement, this study failed to find a similar threshold for UCoF, which agrees with current views on what is “safe.” These results suggest that it may be possible to determine a minimum required shear force at heelstrike for a given movement that, if met by the surface condition, will reduce the probability of a catastrophic slip. Further study is required to support this speculation; however, the approach of studying early stance kinetics shows promise in quantifying slip hazard in a standardized way.

Key Points

A slippery surface causes detectable changes in ground kinetics only 0.42 ms after heelstrike.

Resulting changes in shear force can be detected earlier than for coefficient of friction.

Shear force may be a more promising indicator of slip potential for a surface condition than coefficient of friction.

References

Barnett

R. L.

(2002). “Slip and fall” theory—Extreme order statistics. International Journal of Occupational Safety and Ergonomics: JOSE, 8, 135–159.

Brady

R. A.

Pavol

M. J.

Owings

T. M.

Grabiner

M. D.

(2000). Foot displacement but not velocity predicts the outcome of a slip induced in young subjects while walking. Journal of Biomechanics, 33, 803–808.

Brungraber

R. J.

(1976). An overview of floor slip-resistance research with annotated bibliography (NBS Technical Note 895). Washington, DC: National Bureau of Standards.

Cham

Redfern

M. S.

(2001). Lower extremity corrective reactions to slip events. Journal of Biomechanics, 34, 1439–1445.

Cham

Redfern

M. S.

(2002). Changes in gait when anticipating slippery floors. Gait & Posture, 15, 159–171.

Chang

W. R.

Grönqvist

Leclercq

Myung

Makkonen

Strandberg

Brungraber

R. J.

Mattke

Thorpe

S. C.

(2001). The role of friction in the measurement of slipperiness, Part 1: Friction mechanisms and definition of test conditions. Ergonomics, 44, 1217–1232.

Cooper

R. C.

Prebeau-Menezes

L. M.

Butcher

M. T.

Bertram

J. E. A.

(2008). Step length and required friction in walking. Gait & Posture, 27, 547–551.

DiDomenico

McGorry

R. W.

Chang

C.-C.

(2007). Association of subjective ratings of slipperiness to heel displacement following contact with the floor. Applied Ergonomics, 38, 533–539.

Francis

Zozula

(1990). Experimental determination of limiting and sliding friction forces for purposes of modeling slips, stumbles, and falls. In Gray

(Ed.), Slips, stumbles, and falls: Pedestrian footwear and surfaces (pp. 55–72). West Conshohocken, PA: ASTM International.

10.

Grönqvist

(1999). Slips and falls. In Kumar

(Ed.), Biomechanics in ergonomics (pp. 351–375). London, UK: Taylor & Francis.

11.

Grönqvist

Chang

W. R.

Courtney

T. K.

Leamon

T. B.

Redfern

M. S.

Strandberg

(2001). Measurement of slipperiness: Fundamental concepts and definitions. Ergonomics, 44, 1102–1117.

12.

Grönqvist

Hirvonen

Matz

(2001, July–August). Walking safety and contact time related variation in shoe-floor traction. Paper presented at the International Conference on Computer-Aided Ergonomics and Safety, Slips and Falls Tribology, Maui, HI.

13.

Grönqvist

Hirvonen

Tuusa

(1993). Slipperiness of the shoe-floor interface: Comparison of objective and subjective assessments. Applied Ergonomics, 24, 258–262.

14.

Hanson

J. P.

Redfern

M. S.

Mazumdar

(1999). Predicting slips and falls considering required and available friction. Ergonomics, 42, 1619–1633.

15.

James

D. I.

(1980). A broader look at pedestrian friction. Rubber Chemistry and Technology, 153, 512–541.

16.

Kim

I.-J.

Nagata

(2008). Research on slip resistance measurements—A new challenge. Industrial Health, 46, 66–76.

17.

Leamon

T. B.

K. W.

(1990, September). Microslip length and the perception of slipping. Paper presented at the 23rd International Congress on Occupational Health, Montreal, Canada.

18.

Manning

D. P.

Jones

(2001). The effect of roughness, floor polish, water, oil and ice on underfoot friction: Current safety footwear solings are less slip resistant than microcellular polyurethane. Applied Ergonomics, 32, 185–196.

19.

McGorry

R. W.

DiDomenico

Chang

C.-C.

(2010). The anatomy of a slip: Kinetic and kinematic characteristics of slip and non-slip matched trials. Applied Ergonomics, 41, 41–46.

20.

McVay

E. J.

Redfern

M. S.

(1994). Rampway safety: Foot forces as a function of rampway angle. American Industrial Hygiene Association Journal, 55, 626–634.

21.

Perkins

P. J.

(1978). Measurement of slip between the shoe and ground during walking. In Anderson

Senne

(Eds.), Walkway surfaces: Measurement of slip resistance, ASTM Special Technical Publication 649 (pp. 71–87). Philadelphia, PA: American Society for Testing.

22.

Powers

C. M.

Brault

J. R.

Stefanou

M. A.

Tsai

Y.-J.

Flynn

Siegmund

G. P.

(2007). Assessment of walkway tribometer readings in evaluating slip resistance: A gait-based approach. Journal of Forensic Sciences, 52, 400–405.

23.

Redfern

M. S.

Cham

Gielo-Perczak

Grönqvist

Hirvonen

Lanshammar

Marpet

Pai

C. Y. C.

Powers

(2001). Biomechanics of slips. Ergonomics, 44, 1138–1166.

24.

Redfern

M. S.

DiPasquale

(1997). Biomechanics of descending ramps. Gait and Posture, 6, 119–125.

25.

Redfern

M. S.

Rhoades

T. P.

(1996). Fall prevention in industry using slip resistance testing. In Bhattacharya

McGlothlin

(Eds.), Occupational ergonomics: Theory & applications (pp. 463–476). New York, NY: Marcel Dekker.

26.

Strandberg

(1983). On accident analysis and slip-resistance measurement. Ergonomics, 26, 11–32.

27.

Strandberg

(1985). The effect of conditions underfoot on falling and overexertion accidents. Ergonomics, 28, 131–147.

28.

Strandberg

Lanshammar

(1981). The dynamics of slipping accidents. Journal of Occupational Accidents, 3, 153–162.

29.

Yoon

H.-Y.

Lockhart

T. E.

(2006). Nonfatal occupational injuries associated with slips and falls in the United States. International Journal of Industrial Ergonomics, 36, 83–92.