Date of Submission



This thesis investigated measures of walking instability in older adults with the aim to establish if stride dynamics and gait variability can predict falls in active older women. Falls are a major problem for older adults and the majority occur when walking. The identification of markers of walking instability that predict falls, particularly in active and healthy older adults, would help prevent injury, loss of independence and institutionalisation. Three studies were conducted as part of this thesis. These studies investigated the effect of walking protocol on measures of gait variability in healthy adults and examined the relationship between stride dynamics, gait variability and falls in older adults. Women were used in each study due to their higher incidence of falls and falls-related injury (Stevens & Sogolow, 2005). Study 1 and 2 recorded spatial, temporal and gait variability data in older (age range 57 to 79 years) and younger women (age range 19 to 21 years) screened for conditions that might impact upon balance or walking. Gait data were collected with an 8.1m GAITRite mat for 10 trials of discrete single walks and 10 laps of a continuous circuit, presented in random order. Study 1 examined the test-retest reliability and systematic bias of data recorded during repeated single and continuous over-ground walking trials over two separate test sessions that were seven days apart (median REPLACE2 SD, 7REPLACE21.58d). Paired t tests, intraclass correlation coefficients, standard errors of measurement, and coefficients of variation were calculated. Study 2 investigated if gait variability data captured during repeated single over-ground walking differed from variability data captured during continuous over-ground walking.;To quantify variability, standard deviation and coefficients of variation were calculated for each gait parameter, and paired t tests were used to compare the measures of variability recorded for each walking protocol. A major finding from the first two studies was that gait parameters, including gait variability, differed between walking protocols. Study 1 showed that although both continuous and repeated single walking protocols were reliable, the continuous over-ground walking protocol produced less bias (19% of gait parameters) in test-retest spatiotemporal gait data compared to the single trial condition (65% of gait parameters). Between-protocol differences were more apparent for the older than younger women, with 77% of the gait parameters showing bias in the single trial condition. In contrast, no systematic bias was found in the continuous condition for older adults. In study 2, walking protocol differences were found between the gait variability data. Compared with a continuous over-ground walking protocol, a repeated single over ground protocol resulted in increased variability of velocity, step length and stride length data (p<0.01) for the older women. In the younger women, increased variability of velocity (p?0.02), step length (p=0.04), stride time (p?0.02) and step time (p=0.02) were found for the single walking trials. The findings from studies 1 and 2 suggest that a continuous protocol may be more stable and may detect gait changes more readily, especially for older women. Based on the outcomes of study 1 and 2, gait data recorded using a continuous walking protocol were used for the major analyses in study 3 which examined gait variability and stride dynamics. Additionally, data were also recorded from the equivalent number of repeated single walking trials to investigate the influence of walking protocol upon gait variability and falls.;07 years) underwent screening procedures and completed seven minutes of walking around a continuous circuit. Gait data were collected with an 8.1m GAITRiteREPLACE3 mat and with two tri-axial CrossbowREPLACE3 accelerometers. A small subset (n=12, mean age=67.17 + 5.27) of participants also attended on a second visit one week following their initial testing session to evaluate the test-retest reliability of the accelerometer data. Participants were then followed prospectively for one year to record fall incidence. Differences in physical (demographic and screening), balance, gait variability and stride dynamic measures between fallers (one or more falls) and non-fallers were examined using Multivariate Analyses of Variance (MANOVAs) and independent samples t tests. Between-leg differences in stride dynamics were assessed using a paired samples t test. To evaluate the ability of stride dynamics and gait variability to predict future fallers, direct logistic regression was performed. All analyses were repeated with the sample stratified into non-fallers, single fallers (one fall) and multiple fallers (two or more falls), as well as into multiple fallers and a combined group of single and non-fallers, to investigate the effect of multiple falls as an independent variable. Finally, to evaluate whether walking protocol influenced study outcomes, all between-group and prediction analyses were again repeated using data collected from the repeated single walking protocol. The major finding of this study was that inter-limb dynamics were altered in fallers. Specifically, inter-limb differences (p?0.04) were found in the fractal scaling index of fallers (one or more falls) aged over 70 years, and multiple fallers (two or more falls) aged over 55 years, but not in non-fallers, single fallers or the combined group of single and non-fallers.;05) were found in any physical, balance, gait variability or other stride dynamic measures between those who fell in the subsequent year and those who did not fall. Additionally, no gait variable predicted future falls in the sample of active older women. Similar outcomes were found when data from a repeated single walking protocol were used, and when the sample was stratified in non-fallers, single fallers, multiple fallers and a combined group of single and non-fallers. Therefore, despite no observable difference in other common measures of intrinsic falls risk, control of inter-limb dynamics was reduced in active and otherwise healthy older fallers and multiple fallers. This outcome suggests that inter-limb dynamics could provide a clinically sensitive and possible early detection marker of gait instability and falls risk in high functioning older adults prior to evidence of change in other measures of physical, balance or gait function, including gait variability.

Document Type


Access Rights

Open Access


259 pages

Degree Name

Doctor of Philosophy (PhD)