- Email: [email protected]
The Multiple Tasks Test
Gait and Posture 14 (2001) 191– 202 www.elsevier.com/locate/gaitpost
The Multiple Tasks Test Development and normal strategies Bastiaan R. Bloem a,b,*, Vibeke V. Valkenburg b, Mathilde Slabbekoorn b, Mirjam D. Willemsen b a
Department of Neurology, Uni6ersity Medical Centre, St. Radboud, Nijmegen, The Netherlands b Department of Neurology, Leiden Uni6ersity Medical Centre, Leiden, The Netherlands Received 25 April 2001; accepted 8 May 2001
Abstract Simultaneous challenge of posture and cognition (‘dual tasks’) may predict falls better than tests of isolated components of postural control. We describe a new balance test (the Multiple Tasks Test, MTT) which (1) is based upon simultaneous assessment of multiple ( \2) postural components; (2) represents everyday situations; and (3) can be applied by clinicians. Relevant risk factors for falls and actual fall circumstances (identified from a prospective survey in Parkinson’s disease) were used to design functional tests (or postural ‘components’) that resembled everyday situations. We distinguished a ‘cognitive’ component (answering serial questions) from largely ‘motor’ components (standing up, sitting down, turning around, walking, avoiding obstacles, and touching the floor). Four additional components included carrying an empty or loaded tray, wearing shoes with slippery soles and reduced illumination. These components were combined to yield eight separate tasks of increasing complexity that were executed sequentially. The first and simplest task consisted of standing up, undisturbed walking, turning around and sitting down. For each of the next tasks, a new component was added to the earlier and otherwise identical task. All components within each task had to be performed simultaneously. Errors were defined as Hesitations (slowed performance) or Blocks (complete cessation), which were scored separately for execution of motor and cognitive components. Speed of performance was not stressed, but was measured for all tasks. The MTT was administered to 50 young healthy subjects (mean age 27.6 years) and 13 elderly subjects (mean age 62.0 years). To study learning effects, 20 different young subjects (mean age 21.0 years) received the MTT in order of gradually decreasing complexity. For subjects who received the MTT in order of increasing difficulty, 62% in both age groups performed all eight tasks without any Errors in the motor components. Among those making Errors, the proportion of subjects that made motor Errors increased significantly as the tasks became more complex (F(1,7) = 2.66, P B0.05). This increase differed across the two groups (significant interaction of Group by Task; F(1,7) = 3.07, P= 0.01) because more elderly subjects produced motor Errors during the most complex tasks. Cognitive Errors increased even more than motor Errors with task complexity, and this increase was most pronounced in young subjects (significant interaction of Group by Error Type by Task; F(1,1,7) =3.85, P= 0.001). Only eight young (16%) and four elderly subjects (30.8%) performed the MTT without any motor or cognitive Errors, again suggesting that more young subjects made cognitive Errors. Among subjects who received the MTT in reverse order, motor errors were more common than among subjects who received the MTT in order of increasing complexity (F(1,7) =5.90, PB0.05), particularly during the two most difficult tasks. The elderly performed all tasks slower than the young subjects. We conclude that the MTT is a new balance test based upon a multiple task design that resembles everyday situations. Performance by healthy subjects revealed interesting insights into normal postural strategies. For complex postural tasks, healthy subjects favour execution of motor components over execution of a cognitive component (‘posture first’ strategy). Young subjects were more inclined than elderly subjects to use this strategy. Motor learning influenced performance among subjects who received the MTT in order of increasing difficulty. Further studies must determine whether the MTT can be used to evaluate balance disorders. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Posture; Aging; Dual-task; Falls
* Corresponding author. Present address: Department of Neurology, Radboud Oost, University Medical Centre, St. Radboud, PO Box 9101, 6500 H13 Nijmegen, The Netherlands. Tel.: + 31-24-361-8860; fax: +31-24-354-1122. E-mail address: [email protected] (B.R. Bloem). 0966-6362/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 6 6 - 6 3 6 2 ( 0 1 ) 0 0 1 4 1 - 2
192
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
1. Introduction Falls significantly threaten the quality of life of elderly persons by causing physical injury and serious psychosocial consequences [1]. Mortality is increased after a fall because subjects may suffer deadly falls, or due to underlying diseases and impaired mobility [2]. Not surprisingly, the costs of falls are immense [3]. Therefore, prevention of falls is important. Effective strategies are available [2,4,5], but prevention requires that subjects prone to falls be identified early. Unfortunately, prediction of falls remains difficult, probably because falls are caused by many different and often interrelated factors. Accumulating evidence supports this multifactorial character of postural instability and falls in the elderly. First, various studies showed that falling risks increase with the number of risk factors present [6,7]. Second, falls are best predicted by balance tests that probe this multifactorial nature of postural instability. Thus, rating scales that combine individual test results into a composite score have better predictive value than balance tests that measure isolated aspects (or ‘components’ [8]) of postural control. An example is the Mobility Index, which focuses on different components of posture and gait [9]. This rating scale is among the best predictors of falls [6]. A drawback is that composite scores are ‘post-hoc’ mathematical creations. The individual tests do not challenge multiple components at the same time, but remain focused on separate elements. Truly simultaneous assessment of multiple postural components has been described in experimental settings. Many groups investigated the influence of a secondary task on gait and balance in elderly subjects [10–13] or patients with balance disorders [14–21]. Most investigators focused on gait or balance control during a secondary cognitive task, although some used a secondary motor task [11,15,16,22,23]. A general disadvantage was the focus on balance control in a laboratory setting, rather than a clinically relevant environment. Although no two studies are comparable, the weight of the evidence shows that balance and gait deteriorate when a secondary task is performed. This suggests that even highly practised and seemingly ‘automatic’ processes such as walking require some degree of cognitive processing. The more complex and more novel the postural task, the higher the attentional demands [12,24]. These studies also suggested that ‘dual task’ or truly multiple task performance [25] should be used to predict falls. Lundin-Olsson et al. [26] studied this first. They reported that elderly subjects, who stopped walking when talking had an increased risk of falling. In fact, ‘stops walking when talking’ emerged as one of the best predictors of falls identified so far, particularly for subjects with cognitive impairment. A clear advantage
is the easy applicability by clinicians. A drawback is that the test seems unremarkable in persons without cognitive impairment and may not predict falls caused largely by motor disability (as in Parkinson’s disease) [27]. This may restrict wider practical use. The investigators later showed that difficulty with a secondary manual task (carrying a glass of water) also predicted falls in the elderly [11]. However, most subjects were demented or depressed, leaving unanswered whether falls in cognitively intact persons can be predicted by combining two motor tasks. We wanted to examine the prediction [25] that tests probing multiple (i.e. more than two) postural components would be more sensitive than a strictly dual task design. Interestingly, complex secondary tasks may distinguish better between balance-impaired patients than simpler secondary tasks [15]. Indeed, strictly dual task designs do not always distinguish well between patients and controls, over and above any baseline differences between these groups [27,28]. We speculated that combinations of various motor tasks would be particularly useful for patients without cognitive impairment, because their falls are not predicted by combining a single motor task with a mental task [27]. We further reasoned that falls in daily life would be predicted best by tests that represent complex everyday situations [29]. We also argued that falls would be predicted best by tests that truly challenge postural safety. Finally, we wanted to develop a balance test that would potentially be easy to apply in a consulting room by clinicians. Therefore, our first goal (part A of this paper) was to describe the development of a new balance test that fulfilled the above criteria. Our second goal (part B) was to evaluate normal coping strategies with increasingly complex postural tasks. One possibility is that impaired multiple task performance reflects a limited processing capacity of the central nervous system. If this were correct, healthy subjects should be able to integrate fairly complex postural tasks without errors (although errors would inevitably appear for extremely complex tasks). Alternatively, during complex tasks, healthy subjects might purposely lend priority to complete certain task components at the expense of others [30]. According to this view, a blockade would not reflect pathology but ‘prudent’ behaviour intended to optimise the primary task (maintaining balance). This strategy is termed ‘posture first’ [12]. If such intended ‘priority processes’ exist, subjects should portray slowness or a block in executing certain components of complex postural tasks. Theoretically, the safest postural strategy would be to favour maintaining balance (the ‘primary’ task) over execution of e.g. a manual or mental task. Better insight in these normal strategies is a prerequisite for interpreting pathological processes in balance disorders. To study aging effects, we included both young and elderly subjects.
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
2. Part A: development of a new ‘multiple tasks’ test
2.1. Methods We first identified relevant risk factors for falls from an orienting literature review. Secondly, we identified actual fall circumstances from an earlier prospective survey in Parkinson’s disease [31,32]. These risk factors and fall circumstances were then ‘transformed’ into functional tests (or postural ‘components’) that resembled everyday situations. These components were subsequently combined to yield the MTT.
2.1.1. Literature re6iew We performed an orienting literature review to identify relevant risk factors that could be used to design functional balance tasks. For this purpose, we applied several restrictions to our review. First, retrospective studies were excluded because elderly subjects easily forget falls [33]. Second, we only included analyses of multiple (] 2) falls, injurious falls, or both. Multiple falls are a better index of chronic disorders than single falls, which are often caused by environmental accidents with a low recurrence rate [34] and have little clinical importance, unless injury occurs. Third, because risk factors for falls are often interrelated, we confined the review to risk factors that were independently (in multivariate analyses) associated with falls. Fourth, we restricted our review to risk factors that were consistently (across studies) associated with falls in the elderly. Finally, risk factors were only included if they could be transformed to functional tasks for use in a consulting room. Since our review was orienting, we did not take the methodological quality of the selected papers into account. 2.1.2. Fall circumstances These were obtained from a separate prospective study on falls in 59 Parkinson patients (mean age 61 years; 21 women; mean Hoehn and Yahr score 2.3) and 55 healthy controls (mean age 60 years; 37 women) [31,32]. Subjects recorded the exact circumstances of all falls during 6 months, using standardised scoring forms that were returned directly after each fall. Subjects were also contacted by telephone every 2 weeks to assure that no falls were missed. 2.2. Results 2.2.1. Risk factors for falls Table 1 shows the risk factors independently associated with repeated or injurious falls in the elderly. Several risk factors were patient-related (e.g. visual impairment), whereas others were environmental factors (e.g. poor illumination). Many other risk factors were inconsistently associated with falls. Some risk
193
factors were excluded because they could not be transformed into functional tests (e.g. use of psychotropic medication) or could not be used in consulting rooms (e.g. climbing stairs). We also excluded risk factors that produced tests which were difficult to standardise or score (e.g. ‘abnormal reaction to push or pressure’) [35].
2.2.2. Fall circumstances Patients reported 205 falls, and controls reported 10 falls. Reliable information about fall circumstances was available for 160 falls of patients and all falls of controls (Table 2). 2.2.3. The multiple tasks test (MTT) 2.2.3.1. Task components. Based upon the literature review and fall circumstances, we selected 11 ‘components’ that could be used to design functional tasks for use in a consulting room (Table 3). Several interrelated risk factors or circumstances were combined to yield a single component. For example, we combined the two risk factors ‘poor illumination’ and ‘visual impairment’ into a component where light in the examination room was reduced. Similarly, we combined ‘dizziness upon Table 1 Risk factors identified in the orienting literature review Identified risk factors
Functional components
Poor illuminationa Visual impairmentb Domestic environmenta Dizziness upon standingb Orthostatic hypotensionb Problems rising from a chairb Leg weaknessb
Reduced illumination of test environment Reduced illumination of test environment Living room sitting; obstacles on floor Standing up from a chair Standing up from a chair Standing up from a chair
Standing up from a chair; squatting and touching the floor Lower extremity Standing up from a chair; squatting and disabilityb touching the floor Gait impairmentb Undisturbed walking; avoiding obstacles Stops walking when Performance of a simultaneous mental task talkingb Cognitive Performance of a simultaneous mental task impairment/demen tiab Turningb Turning 180° Presence of multiple Combinations of the above tests risk factorsa,b The second column shows the functional ‘components’, into which these risk factors could be transformed. Examples of excluded factors included (psychotropic) medication, climbing stairs, dressing impairment, reaching, decreased neck range of motion and abnormal reaction to push or pressure. a Environmental factors. b Patient-related factors.
194
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Table 2 Fall circumstances in moderately affected patients with Parkinson’s disease Fall circumstances
Functional components
Walkinga Obstacles on the floorsb Wearing shoes without proper gripb Slippery floorb Standing upa Carrying objects in the handsa,b
Undisturbed walking Avoiding obstacles Shoes with a slippery sole
Squatting, bending downb While turning arounda Doing at least two things simultaneouslya,c
Shoes with a slippery sole Standing up from a chair Carrying an empty or loaded tray Squatting and touching the floor Turning around 180° Combinations of the above components
The second column shows the functional ‘components’, into which these fall circumstances could be transformed. Examples of excluded circumstances included reaching, climbing stairs and freezing. a Patient-related factors. b Environmental factors. c This occurred in 79 out of 160 falls (49.4%) in Parkinson patients.
standing’, ‘orthostasis’ and ‘inability to rise from a chair’ into a component where subjects were asked to stand up from a chair. Different types of components could be distinguished. The first type was a cognitive component that consisted of a continuous mental task. The second type consisted of components that (largely) challenged the Table 3 Components selected for use in the Multiple Tasks Test are shown in the first column, while the respective tasks are shown in the top row
The table also shows which components were used (indicated by a ‘+’ sign) during each of the eight consecutive tasks. The shaded areas index the components that were used for scoring purposes.
motor system. This included standing up from a chair, undisturbed walking, carrying an empty tray, squatting and touching the floor, turning around and sitting down on a chair. Some components also demanded particular attention, e.g. carrying a loaded tray and avoiding obstacles on the floor. Carrying a loaded tray resembled the tasks (carrying a glass of water or a tray with glasses) used by others [11,15]. The third type was a visual component (reduced illumination in the room). The final component consisted of wearing shoes with slippery soles. These slippery shoes were promising because confrontation with new footwear may unveil multiple task difficulties that would otherwise remain unnoticed due to compensatory mechanisms, particularly in patients with longstanding disease and gradually developing lesions [18]. It was difficult to choose a proper cognitive task that continuously challenged mental processes. During pilot studies, we used open questions (e.g. ‘name as many trees as possible’). However, such questions had a disproportionate influence on task performance, as even young controls frequently blocked all movements (including answering). Conversely, asking subjects to produce serial numbers (e.g. 3-6-9-12) proved too simple and was seemingly processed automatically. Others also found that relatively simple arithmetic tasks produced little interference with concurrent, competing tasks [20]. Counting backward (e.g. serial sevens, starting from 100) produced considerable inter-individual differences. We, therefore, chose for a continuous series of relatively simple questions regarding everyday situations (e.g. ‘What is the date?’ or ‘What did you have for breakfast this morning?’). The examiner walking besides the subject posed each next question (from a standard list of 150 different questions) directly after the answer to the earlier question was given. This provided a continuous mental challenge that produced comparable results among a homogenous group of young controls. Given the difficulties in selecting an appropriate secondary cognitive task, we did not attempt to include a second, different cognitive task. Note also in this respect that Haggard et al. [20] recently failed to find relevant differences between four very different secondary cognitive tasks.
2.2.3.2. Functional tasks. The 11 components were combined to yield eight sequential tasks of increasing difficulty, due to simultaneous challenge of an increasing number of components. The MTT was performed in a quiet room (8× 3 m, linoleum floor) that was adequately illuminated. A chair was placed at each end of the room. Three obstacles (two were 9 cm wide and 3 cm high, one was 36 cm wide and 1.5 cm high) were positioned on the floor at variable distances (between 1 and 2 m). Performance was recorded on videotape.
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Undisturbed walking and ‘stops walking when talking’, which had been tested during baseline examination, were uninformative in our subjects. We, therefore, made the first task slightly more complicated by asking subjects to stand up from a chair, walk undisturbed along a predefined course, turn 180° and sit down again. This task was repeated seven times, but each time an extra component was added to the earlier and otherwise identical task (Table 3). During the second task, subjects answered a continuous series of brief questions while walking. During the third task, subjects avoided the obstacles on the floor. During the fourth task, subjects carried an empty tray. During the fifth task, the tray was loaded with two hardboiled eggs in cups and one loosely rolling egg. During the sixth task, subjects wore indoor shoes with slippery soles. During the seventh task, subjects squatted and tipped the floor halfway the obstacle course. During the eighth task, subjects wore sunglasses, while illumination was moderately reduced. Use of sunglasses allowed us to leave the room sufficiently lit for videotape recording. Subjects were instructed not to prioritise any given component, but to combine all components of each task as good as possible, at their own preferred speed. Most motor components could be executed simultaneously, although some were in fact executed directly after each other, such as touching the floor while walking, or sitting down after walking. Unlike some other studies [22], we urged subjects only once (at the beginning of the experiment) to not purposely prioritise any given component. If this instruction is continuously repeated, one might theoretically obscure any tendency to ‘disobey’ the initial instruction and to lend priority to what subjects perceive as the primary task (e.g. maintaining balance). Study of such priority strategies was a main goal of our study. During all tasks, the investigator walked beside the subject to prevent falls. We considered using a safety harness attached to a low-friction overhead track, but rejected this because it would hamper practical use in a consulting room.
2.2.3.3. Scoring system. Scoring was partially based upon subjective (qualitative) interpretation of subjects’ performance on separate task components. The simplicity of this approach has clear advantages for use in a clinical setting. Moreover, quantitative scoring during the test (e.g. with a stopwatch) would be unpractical because many test components had to be executed (and scored) continuously. Four components (carrying the unloaded or loaded tray, wearing slippery shoes and reduced illumination) could not be scored independently, but served to complicate the task and thus facilitate production of Errors. For scoring purposes, the other components were divided
195
into ‘motor’ (standing up, sitting down, walking, avoiding obstacles, turning around, touching the floor) and ‘cognitive’ components (answering questions). Impaired multiple task performance can be reflected by slowing [11,17,23] or a complete stop [26,36] in executing one or more components. Therefore, all tasks were scored as follows: rapid performance of all components within the task (‘Normal’); obvious slowing in one or more components within the task (‘Hesitation’); complete stop or inability to perform one or more components within the task (‘Block’). For example, a Hesitation was scored if subjects clearly answered the questions slower than their own baseline answering performance, which was first determined for each individual while seated. This baseline performance always produced direct answers to the simple questions without any hesitation in seated subjects, which were screened to exclude disorders that might affect the ability to answer, such as impaired cognition or dysarthria. Compared with this baseline performance, hesitations (delays) in answering were always very obvious during the multitasking conditions (silence for at least 1 s, or ‘uh-uh’ sounds). A Block was scored if subjects entirely stopped answering questions. Similarly, performance during walking, standing up, turning and sitting down was compared with baseline performance during the Tinetti Mobility Index. Hesitations and Blocks will be analysed separately, but will also jointly be referred to as ‘Errors’. The score was determined for all eight consecutive tasks of the MTT. Since we were interested in individual performance, our scoring system produced the proportion of subjects with either a completely error-free performance, as well as the proportion of subjects that made at least one Error during any given test. Hence, subjects received an abnormal test score if they made at least one Error (Hesitation or Block) during a given task. Conversely, subjects only received a normal score if they performed all components within a given task without any Error. Absolute numbers of Errors (Hesitations or Bocks) for each task were not scored because scoring individual performance is more helpful from a clinical perspective as a diagnostic tool. Scoring was done directly during the tasks. In addition to the above-described qualitative scoring, we also quantified movement time objectively using a stopwatch. For this purpose, performance was recorded on videotape. Although subjects were left free to execute the tasks at their own preferred speed, we used these videotapes to quantify the time between start (standing up) and end of each task (seated position) as an extra outcome variable.
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
196
Table 4 Performances for motor components within each of the eight MTT tasks Task
Young controls (N= 50) N
One Two Three Four Five Six Seven Eight
50 46 46 50 47 41 47 47
H (100) (92) (92) (100) (94) (82) (94) (94)
0 2 2 0 3 9 3 3
B (0) (4) (4) (0) (6) (18) (6) (6)
P-value ( 2)
Elderly controls (N = 13)
0 2 2 0 0 0 0 0
(0) (4) (4) (0) (0) (0) (0) (0)
N
H
13 (100) 12 (92) 12 (92) 11 (84) 10 (77) 12 (92) 9 (70) 13(100)
0 0 0 1 1 1 2 0
B (0) (0) (0) (8) (8) (8) (15) (0)
0 1 1 1 2 0 2 0
(0) (8) (8) (8) (15) (0) (15) (0)
0.67 0.67 0.02 0.02 0.34 B0.01 0.49
The numbers of subjects are shown (percentage between parentheses) with a normal, rapid performance (N), a motor Hesitation (H) or a motor Block (B). Hesitations or Blocks in the cognitive component (answering serial questions) were ignored for this analysis. The P-values refer to differences in performance between the two groups.
3. Part B: normal coping strategies for complex postural tasks
3.1. Subjects and methods Experiment 1. Fifty young subjects (29 women, mean ( 9 S.D.) age 27.69 6.6 years) and 13 elderly subjects (six women, mean age 62.097.8 years) received the MTT in order of increasing difficulty. Experiment 2. Twenty different healthy young subjects (eight women, mean age 20.192.2 years) received the MTT in order of decreasing difficulty. We used fairly strict inclusion criteria to select a rather homogeneous sample of elderly persons without any obvious physical or cognitive problems. Without strict inclusion criteria, elderly persons with underlying, partially subclinical age-related diseases might be included as well, and this could increase the variability in the data [37– 39]. Thus, for both experiments, history taking and detailed physical examination (including the Tinetti Mobility Index [9]) were used to exclude balance problems and neurological, orthopaedic, speech or visual disorders in all subjects. The Mini Mental State Examination [40] was used to exclude cognitive problems in the elderly subjects (mean score was 29.0; range 25– 30). All subjects gave informed consent as approved by the Ethical Committee of the Leiden University Medical Centre.
3.1.1. Statistical analyses A two-way (group by task complexity) MANOVA for repeated measures was used to compare the number of subjects who produced Errors (i.e. Hesitations or Blocks) for each task across young and elderly subjects. This was done separately for motor Errors and for cognitive Errors. In addition, to evaluate if cognitive Errors changed differently than motor Errors with task complexity, a three-way (error type by group by task complexity) MANOVA for repeated measures was
used. Greenhouse–Geisser Epsilon was used to correct for non-sphericity. These analyses were used after we ascertained that identical results were obtained when the data were fitted using a random-effects Poisson regression model. In addition, we compared the proportions of subjects who made Errors for each individual task using the Chi-square test ( 2-test). A similar analysis using a two-way (sequence direction by task complexity) MANOVA for repeated measures was used to compare the number of subjects who produced Errors for each task across subjects who received the MTT in order of increasing difficulty and subjects who received the MTT in reverse order. The log-rank test was used to study whether the number of subjects that performed all eight tasks without Errors differed between young and elderly subjects, and between subjects who received the MTT in order of increasing difficulty versus subjects who received the MTT in reverse order. Relative risks (and 95% confidence intervals) of making an Error in at least one component of the test were calculated using a Cox-proportional hazards model. Finally, the time taken to complete each task was compared between young and elderly subjects using a two-way (group by task complexity) MANOVA for repeated measures, followed by post-hoc comparisons using Tukey’s test to correct for multiple comparisons.
3.2. Results 3.2.1. The MTT in young and elderly healthy subjects Table 4 shows the performance of motor components. All subjects completed the MTT without falling. Four young subjects made Errors during the second task. Two of them had a Block (stopped walking). Most motor Hesitations occurred for the sixth task (when subjects wore slippery shoes for the first time). Overall, the proportion of subjects that produced Hesitations or Blocks for the motor components differed between young and elderly subjects (F(1,7)= 5.12, PB
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
0.05), more elderly subjects making Errors than young subjects. Table 4 shows that the proportion of subjects with motor Hesitations or Blocks was higher among elderly controls during the fourth, fifth and particularly the seventh task. Furthermore, the proportion of subjects that made motor Errors increased significantly as the tasks became more complex (F(1,7) = 2.66, PB 0.05). This increase differed across the two groups (significant interaction of Task by Group; F(1,7)= 3.07, P =0.01) because particularly elderly controls produced motor Errors during the more complex tasks. In all 62% of subjects in both groups performed all eight tasks without any motor Error (Fig. 1). Note that Fig. 1 provides complementary information to Table 4, which shows performance for all subjects for each task. In contrast, the survival analysis presented in Fig. 1 implies that anyone who produced an Error during a given task did not proceed to the next task. Different results were obtained for the cognitive Hesitations or Blocks. With task complexity, the proportion of subjects making cognitive Errors increased even more than for motor Errors, and this increase was now most pronounced in young subjects (significant interaction of Group by Error Type by Task; F(1,1,7) =3.85,
197
P= 0.001). Fig. 2 shows the performance when Hesitations or Blocks were scored for both the cognitive and motor components. Only 16.0% of young controls completed all eight tasks without any Error, due to an increasing number of different subjects with cognitive Errors as task complexity increased (note again the difference between this survival analysis and the group results of Table 4). The strategy of young subjects apparently was to postpone answering until the motor components had been completed safely. The elderly performed somewhat better than young subjects because more elderly subjects (30.8%) performed all eight tasks without any motor or cognitive Errors (no significant difference).
3.2.2. Mo6ement time Movement time increased significantly as the tasks became more complex (F(1,7)= 187.79, PB0.001). This increase in movement time differed across both groups (significant interaction of Group by Task; F(1,7)= 5.02, P= 0.001) because movement time increased more steeply in elderly subjects than young subjects. The total time to complete the MTT differed between both groups (F(1,7)= 10.79, PB 0.005) be-
Fig. 1. Kaplan – Meier curves for the cumulative proportion of subjects with a completely Error-free performance for all motor components within each respective task of the MTT. Subjects who made an Error (Hesitation or Block) for at least one motor component of any given task were excluded from the following tasks. Errors in the cognitive component (answering serial questions) were ignored for this analysis. In both groups, 62.0% of subjects had an Error-free performance.
198
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Fig. 2. Kaplan – Meier curves for the cumulative proportion of subjects with a completely Error-free performance for all components (both motor and cognitive) within each respective task of the MTT. Subjects who made an Error for at least one component of any given task were excluded from the following tasks. Sixteen percent of the young controls and 30.8% of the elderly controls completed the MTT without any Errors (no significant difference).
cause elderly subjects performed all eight tasks slower than young controls. Particularly the time to complete the eighth and most difficult task was increased in elderly subjects (mean 23.8 s, range 15.7– 34.1) compared with young subjects (mean 18.5 s, range 10.8– 26.2; P B0.05).
3.2.3. The MTT in re6erse order Motor Errors were more common among subjects who received the MTT in reverse order than among subjects who received the MTT in order of increasing difficulty (F(1,7)= 5.90, P B0.05). The proportion of subjects making motor Errors increased as the tasks became more complex in both groups (F(1,7) =3.25, P B 0.01), and this increase differed across the two groups (significant interaction of Sequence by Task; F(1,7)= 2.96, P =0.01). Particularly the two most difficult tasks produced more subjects making motor Errors among those who directly received them at the beginning of the experiment (Table 5). Conversely, virtually no subjects made Errors during the two simplest tasks in both groups. This is differently illustrated (survival analysis) in Fig. 3, which shows that only 40% of subjects made no motor Errors for the reverse MTT,
as opposed to 62% for those who received the tasks in order of increasing difficulty. Compared with subjects who received the tasks in order of increasing difficulty, the relative risk of making an Error in at least one motor component of the reverse MTT was 2.45 (95% confidence interval 1.18–5.08; PB 0.05).
4. Discussion
4.1. Strategies in healthy subjects This study shows that healthy subjects manifest Hesitations or Blocks while performing complex tasks. This occurred not only in elderly subjects, but also in young subjects. Since all subjects had a normal balance, this suggests that a normal strategy is to lend priority to complete certain aspects of a complex task, at the expense of others. As task complexity increased, more subjects made motor and cognitive Errors. Complete Blocks occurred less often than Hesitations, and no subject fell. These observations suggest that Errors in multitasking are not necessarily a marker of postural instability or pathologically impaired central processing
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
capacity. Instead, in healthy subjects, Hesitations or Blocks made during a complex task may reflect ‘prudent’ behaviour, designed to optimise postural control and to avoid falls.
4.2. Influence of ageing The MTT revealed some differences between young and elderly subjects. More elderly subjects made motor Errors, in particular during some of the most complex tasks. In contrast, fewer elderly subjects seemed to make cognitive Errors during the MTT. This suggests that young subjects were perhaps more inclined to use a ‘posture first’ strategy than elderly subjects. Others also noted that particularly young subjects favour execution of the postural task at the expense of cognitive errors [22,24,29]. The ‘posture first’ strategy is chosen particularly if the postural task is perceived as hazardous [12,24], as likely occurred during the MTT. Similar priorities are made under many everyday circumstances, e.g. car drivers who cease talking while approaching a hazardous crossing [30]. We also found that elderly subjects performed all tasks somewhat slower than younger subjects. Young subjects possibly paid the price for their ‘speedy’ performance by making more cognitive Errors during complex tasks than elderly subjects. Overall, the differences between young and old subjects were relatively subtle. One explanation is that our elderly subjects were younger (mean age 62 years) than those studied by e.g. Lundin-Olsson et al. [26] (mean age 80 years), Woollacott et al. [41] (mean age 79 years), Brown et al. [42] (mean age 79 years) and Shumway-Cook et al. [12] (mean age 74 years). Another reason is that we studied relatively few elderly subjects, certainly compared with the much larger group of young controls. Variability in performance increases
199
with aging, and larger sample sizes might provide better insight into the range of this variability and perhaps their underlying causes. However, there is mounting evidence that this variability is not primarily caused by aging itself, but rather by underlying (and perhaps otherwise subclinical) diseases [37–39]. Variability is considerably less in carefully selected elderly subjects [43]. We tried to study aging itself by including a rather homogeneous sample of carefully selected elderly persons without obvious physical or cognitive problems upon clinical examination. Indeed, none of our elderly subjects had balance impairment or cognitive deterioration, and performance was rather consistent across elderly persons. This careful selection probably also explains the modest differences with the younger subjects. We do not believe that increasing the number of elderly controls would have led to fundamentally different insights. However, it will be important to perform future studies in less healthy old people to further unravel the contributions of aging and disease to multiple task performance. Impaired multiple task performance likely reveals more abnormalities in patients with balance disorders or cognitive deficits. Indeed, our first experience indicates that the MTT can clearly distinguish patients with Parkinson’s disease from healthy elderly controls [44].
4.3. Learning effects It is possible that motor learning (performance gain through practice) influenced the first experiment, where subjects consistently received tasks identical to earlier ones, except for one novel component. We, therefore, performed a second experiment, where young subjects received the MTT in reverse order (i.e. the most difficult task first, while successive components were eliminated for each of the next tasks). We reasoned that
Table 5 Performances in subjects who received the MTT in order of increasing difficulty versus subjects who received the eight tasks in order of decreasing difficulty Task
Increasing difficulty (N= 50) N
Motor errors One Two Three Four Five Six Seven Eight
50 46 46 50 47 41 47 47
E
(100) (92) (92) (100) (94) (82) (94) (94)
P-value ( 2)
Decreasing difficulty (N =20)
0 4 4 0 3 9 3 3
N
(0) (8) (8) (0) (6) (18) (6) (6)
20 20 18 18 16 17 14 14
E
(100) (100) (90) (90) (80) (85) (70) (70)
0 0 2 2 4 3 6 6
(0) (0) (10) (10) (20) (15) (30) (30)
– 0.43 0.43 0.08 0.06 0.21 0.02 0.01
Subjects with a Hesitation (H) or a Block (B) are shown together as Errors (E). The remaining subjects had a normal, rapid performance (N). Numbers of subjects (percentage between parentheses) are shown. Only Errors for the motor components are shown in this table, but a similar pattern emerged when both motor and cognitive Errors were scored. The P-values refer to differences in performance between the two groups.
200
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Fig. 3. Comparison between 20 young subjects who received the MTT in order of decreasing difficulty (most difficult task first), as opposed to the 50 young subjects described earlier who received the MTT in order of increasing difficulty (simplest task first).
if learning effects were present, the most difficult task would produce more Errors in subjects who directly received this task without carry-over from earlier, less complex tasks. This was indeed observed. Subjects who received the most difficult tasks first made more Errors than subjects who received these tasks at the end. We, therefore, conclude that learning explained at least part of the results during the first experiment, at least in the younger subjects.
4.4. The Multiple Tasks Test: methodological aspects As reviewed in Section 1, others also investigated the influence of secondary tasks on gait and balance in elderly subjects and patients with a variety of diseases. Similar to our findings, these studies showed that dual task performance negatively affects gait and balance. However, our study is one of the first to examine the effect of multiple (more than two) tasks on balance and gait. An advantage of our approach is that the MTT is based upon complex situations that subjects may encounter in daily life. One might argue that the most complex tasks no longer resembled everyday situations. However, our analysis of fall circumstances showed that these tasks were not as far-fetched as they may initially seem [32]. Thus, falls in Parkinson patients typically occurred under fairly complex situations (e.g. carrying a loaded tray into dimly lit rooms with doorsteps and other obstacles, wearing inappropriate
footwear and while talking to a partner). An advantage of the difficulties experienced by healthy subjects is that the MTT produces measurable results even in control groups. This may prove beneficial for its use as a diagnostic tool. Geurts et al. [29] suggested that a multiple task design should contain the following components: perceptual manipulations (e.g. distorted visual information), cogniti6e manipulations, motor manipulations (e.g. turning) and mechanical manipulations (e.g. avoiding obstacles). These manipulations should be combined to produce complex environmental conditions. The MTT fulfilled all these requirements. In contrast to others [11], we did not emphasise speed of performance, again because we wanted to replicate daily life situations. Under normal circumstances, elderly subjects are likely to trade off velocity for safety and adopt a slower and more secure performance [45]. Yet, movement time can still be measured as an extra outcome variable, and this indeed distinguished young from elderly subjects. An important goal was to study concepts underlying multiple task performance, and the current MTT served that purpose. Of course, in its present form the MTT is not an ‘end-product’ and has shortcomings. One drawback is the subjective scoring system. While advantageous for clinical use in a consulting room (no complicated equipment is necessary), it is potentially subject to individual bias. Using the videotapes, we are
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
now investigating intra- and inter-observer reliability. Another drawback is the duration of the complete MTT, although even elderly subjects required only a few minutes to complete it. Furthermore, the sequence in which the respective components were added to the tasks was only one of many possibilities. However, administering the separate components in a random sequence would have required a much larger sample size to obtain statistically meaningful results for each test sequence. We hope to simplify the MTT, e.g. by evaluating which components are most informative in balance impaired patients. Perhaps balance problems might be identified earlier when subjects first receive more complex tasks. This would obviate the necessity for less complex tasks and shorten the test considerably.
Acknowledgements G. van der Giessen is gratefully acknowledged for his expert assistance. We thank Dr J.G. van Dijk and Dr D.J. Beckley for their critical comments, and Dr A.H. Zwinderman for his statistical analyses.
References [1] Maki BE, Holliday PJ, Topper AK. Fear of falling and postural performance in the elderly. J Gerontol 1991;45:M123 –31. [2] Bloem BR, van Vugt JPP, Beckley DJ. Postural instability and falls in Parkinson’s disease, Adv Neurol 2001;87:209 –23. [3] Sattin RW. Falls among older persons: a public health perspective. Annu Rev Publ Health 1992;13:489 – 508. [4] Banks MA, Caird FI. Physiotherapy benefits patients with Parkinson’s disease. Clin Rehab 1989;3:11 –6. [5] Tinetti ME, Baker DI, McAvay G, et al. A multifactorial intervention to reduce the risk of falling among elderly people living in the community. New Engl J Med 1994;331:821 – 7. [6] Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. New Engl J Med 1988;319:1701 – 7. [7] Tromp AM, Smit JH, Deeg DJH, et al. Predictors for falls and fractures in the longitudinal aging study Amsterdam. J Bone Miner Res 1998;13:1932 –9. [8] Horak FB, Shupert CL, Mirka A. Components of postural dyscontrol in the elderly: a review. Neurobiol Aging 1989;10:727 – 38. [9] Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc 1986;34:119 –26. [10] Maylor EA, Wing AM. Age differences in postural stability are increased by additional cognitive demands. J Gerontol Psychol Sci 1996;51B:P143 – 54. [11] Lundin-Olsson L, Nyberg L, Gustafson Y. Attention, frailty, and falls: the effect of a manual task on basic mobility. J Am Geriatr Soc 1998;46:758 –61. [12] Shumway-Cook A, Woollacott MH, Kerns KA, Baldwin M. The effects of two types of cognitive tasks on postural stability in older adults with and without a history of falls. J Gerontol Med Sci 1997;52A:M232 –40.
201
[13] Marsh AP, Geel SE. The effect of age on the attentional demands of postural control. Gait Posture 2000;12:105 –13. [14] Camicioli RM, Oken BS, Sexton G, et al. Verbal fluency task affects gait in Parkinson’s disease with motor freezing. J Geriatr Psychiatry Neurol 1998;11:181 – 5. [15] Bond JM, Morris M. Goal-directed secondary motor tasks: their effects on gait in subjects with Parkinson disease. Arch Phys Med Rehab 2000;81:110 – 6. [16] Alexander NB, Mollo JM, Giordani B, et al. Maintenance of balance, gait patterns, and obstacle clearance in Alzheimer’s disease. Neurology 1995;45:908 – 14. [17] Camicioli RM, Howieson DB, Lehman S, Kaye J. Talking while walking. The effect of a dual task in aging and Alzheimer’s disease. Neurology 1997;48:955 – 8. [18] Geurts AC, Mulder TW, Nienhuis B, Rijken RA. Influence of orthopaedic footwear on postural control in patients with hereditary motor and sensory neuropathy. Arch Phys Med Rehab 1992;73:569 – 72. [19] Courtemanche R, Teasdale N, Boucher P, et al. Gait problems in diabetic neuropathic patients. Arch Phys Med Rehab 1996;77:849 – 55. [20] Haggard P, Cockburn J, Cock J, et al. Interference between gait and cognitive tasks in a rehabilitating neurological population. J Neurol Neurosurg Psychiatry 2000;69:479 – 86. [21] Andersson G, Yardley L, Luxon L. A dual-task study of interference between mental activity and control of balance. Am J Otol 1998;19:632 – 7. [22] Chen HC, Schultz AB, Ashton-Miller JA, et al. Stepping over obstacles: dividing attention impairs performance of old more than young adults. J Gerontol Med Sci 1996;51:M116 – 22. [23] Means KM, Rodell DE, O’Sullivan PS. Obstacle course performance and risk of falling in community-dwelling elderly persons. Arch Phys Med Rehab 1998;79:1570 – 6. [24] Lajoie Y, Teasdale N, Bard C, Fleury M. Attentional demands for static and dynamic equilibrium. Exp Brain Res 1993;97:139 – 44. [25] Mulder TW, Berndt H, Pauwels J, Nienhuis B. Sensorimotor adaptability in the elderly and disabled. In: Stelmach GE, Ho¨ mberg V, editors. Sensorimotor Impairment in the Elderly. Amsterdam: Kluwer, 1993:413 – 26. [26] Lundin-Olsson L, Nyberg L, Gustafson Y. Stops walking when talking as a predictor of falls in elderly people. Lancet 1997;349:617. [27] Bloem BR, Grimbergen YAM, Cramer M, Valkenburg VV. Stops walking when talking does not predict falls in Parkinson’s disease. Ann Neurol 2000;48:268. [28] Morris ME, Iansek R, Smithson F, Huxham F. Postural instability in Parkinson’s disease: a comparison with and without a concurrent task. Gait Posture 2000;12:205 – 16. [29] Geurts AC, Mulder TW, Nienhuis B. From the analysis of movements to the analysis of skills. J Rehab Sci 1991;4:9 –12. [30] Mulder TW, Geurts AC. The assessment of motor dysfunctions: preliminaries to a disability-oriented approach. Human Movement Sci 1991;10:565 – 74. [31] Bloem BR, Grimbergen YAM, Cramer M, Willemsen M, Zwinderman AH. Prospective assessment of falls in Parkinson’s disease. J Neurol 2001; in press. [32] Willemsen MD, Grimbergen YAM, Slabbekoorn M, Bloem BR. Vallen bij de ziekte van Parkinson: vaker door houdingsinstabiliteit dan door omgevingsfactoren. Ned Tijdschr Geneeskd 2000;144:2309 – 14. [33] Cummings SR, Nevitt MC, Kidd S. Forgetting falls. The limited accuracy of recall of falls in the elderly. J Am Geriatr Soc 1988;36:613 – 6. [34] Nevitt MC, Cummings SR, Kidd S, Black D. Risk factors for
202
[35]
[36]
[37] [38]
[39]
[40]
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202 recurrent nonsyncopal falls. A prospective study. J Am Med Assoc 1989;261:2663 –8. Bloem BR, Beckley DJ, van Hilten JJ, Roos RAC. Clinimetrics of postural instability in Parkinson’s disease. J Neurol 1998;245:669 – 73. Morris ME, Iansek R, Matyas TA, Summers JJ. Stride length regulation in Parkinson’s disease. Normalization strategies and underlying mechanisms. Brain 1996;119:551 –68. Rowe JW, Kahn RL. Human aging: usual and successful. Science 1987;237:143 – 9. Howieson DB, Holm LA, Kaye JA, et al. Neurologic function in the optimally healthy oldest old: neuropsychological evaluation. Neurology 1993;43:1882 –6. Bloem BR, Gussekloo J, Lagaay AM, et al. Idiopathic senile gait disorders are signs of subclinical disease. J Am Geriatr Soc 2000;48:1098 – 101. Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A
[41]
[42]
[43] [44]
[45]
practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189 – 98. Woollacott MH, Shumway-Cook A. Attentional demands in postural tasks: changes in both healthy and balance impaired adults. Gait Posture 1999;9(Suppl 1):S12. Brown LA, Shumway-Cook A, Woollacott MH. Attentional demands and postural recovery: the effects of aging. J Gerontol Med Sci 1999;54A:M165 – 71. Calne DB, Eisen A, Meneilly A. Normal aging of the nervous system. Ann Neurol 1991;29:206 – 7. Bloem BR, Valkenburg VV, Slabbekoorn M, Van Dijk JG. The Multiple Tasks test. Strategies in Parkinson’s disease. Exp Brain Res 2001;137:478 – 96. Patla AE. A framework for understanding mobility problems in the elderly. In: Craik RL, Oatis CA, editors. Gait Analysis. Theory and Application. St. Louis: Mosby Year Book, 1994:436 – 49.
The Multiple Tasks Test Development and normal strategies Bastiaan R. Bloem a,b,*, Vibeke V. Valkenburg b, Mathilde Slabbekoorn b, Mirjam D. Willemsen b a
Department of Neurology, Uni6ersity Medical Centre, St. Radboud, Nijmegen, The Netherlands b Department of Neurology, Leiden Uni6ersity Medical Centre, Leiden, The Netherlands Received 25 April 2001; accepted 8 May 2001
Abstract Simultaneous challenge of posture and cognition (‘dual tasks’) may predict falls better than tests of isolated components of postural control. We describe a new balance test (the Multiple Tasks Test, MTT) which (1) is based upon simultaneous assessment of multiple ( \2) postural components; (2) represents everyday situations; and (3) can be applied by clinicians. Relevant risk factors for falls and actual fall circumstances (identified from a prospective survey in Parkinson’s disease) were used to design functional tests (or postural ‘components’) that resembled everyday situations. We distinguished a ‘cognitive’ component (answering serial questions) from largely ‘motor’ components (standing up, sitting down, turning around, walking, avoiding obstacles, and touching the floor). Four additional components included carrying an empty or loaded tray, wearing shoes with slippery soles and reduced illumination. These components were combined to yield eight separate tasks of increasing complexity that were executed sequentially. The first and simplest task consisted of standing up, undisturbed walking, turning around and sitting down. For each of the next tasks, a new component was added to the earlier and otherwise identical task. All components within each task had to be performed simultaneously. Errors were defined as Hesitations (slowed performance) or Blocks (complete cessation), which were scored separately for execution of motor and cognitive components. Speed of performance was not stressed, but was measured for all tasks. The MTT was administered to 50 young healthy subjects (mean age 27.6 years) and 13 elderly subjects (mean age 62.0 years). To study learning effects, 20 different young subjects (mean age 21.0 years) received the MTT in order of gradually decreasing complexity. For subjects who received the MTT in order of increasing difficulty, 62% in both age groups performed all eight tasks without any Errors in the motor components. Among those making Errors, the proportion of subjects that made motor Errors increased significantly as the tasks became more complex (F(1,7) = 2.66, P B0.05). This increase differed across the two groups (significant interaction of Group by Task; F(1,7) = 3.07, P= 0.01) because more elderly subjects produced motor Errors during the most complex tasks. Cognitive Errors increased even more than motor Errors with task complexity, and this increase was most pronounced in young subjects (significant interaction of Group by Error Type by Task; F(1,1,7) =3.85, P= 0.001). Only eight young (16%) and four elderly subjects (30.8%) performed the MTT without any motor or cognitive Errors, again suggesting that more young subjects made cognitive Errors. Among subjects who received the MTT in reverse order, motor errors were more common than among subjects who received the MTT in order of increasing complexity (F(1,7) =5.90, PB0.05), particularly during the two most difficult tasks. The elderly performed all tasks slower than the young subjects. We conclude that the MTT is a new balance test based upon a multiple task design that resembles everyday situations. Performance by healthy subjects revealed interesting insights into normal postural strategies. For complex postural tasks, healthy subjects favour execution of motor components over execution of a cognitive component (‘posture first’ strategy). Young subjects were more inclined than elderly subjects to use this strategy. Motor learning influenced performance among subjects who received the MTT in order of increasing difficulty. Further studies must determine whether the MTT can be used to evaluate balance disorders. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Posture; Aging; Dual-task; Falls
* Corresponding author. Present address: Department of Neurology, Radboud Oost, University Medical Centre, St. Radboud, PO Box 9101, 6500 H13 Nijmegen, The Netherlands. Tel.: + 31-24-361-8860; fax: +31-24-354-1122. E-mail address: [email protected] (B.R. Bloem). 0966-6362/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 6 6 - 6 3 6 2 ( 0 1 ) 0 0 1 4 1 - 2
192
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
1. Introduction Falls significantly threaten the quality of life of elderly persons by causing physical injury and serious psychosocial consequences [1]. Mortality is increased after a fall because subjects may suffer deadly falls, or due to underlying diseases and impaired mobility [2]. Not surprisingly, the costs of falls are immense [3]. Therefore, prevention of falls is important. Effective strategies are available [2,4,5], but prevention requires that subjects prone to falls be identified early. Unfortunately, prediction of falls remains difficult, probably because falls are caused by many different and often interrelated factors. Accumulating evidence supports this multifactorial character of postural instability and falls in the elderly. First, various studies showed that falling risks increase with the number of risk factors present [6,7]. Second, falls are best predicted by balance tests that probe this multifactorial nature of postural instability. Thus, rating scales that combine individual test results into a composite score have better predictive value than balance tests that measure isolated aspects (or ‘components’ [8]) of postural control. An example is the Mobility Index, which focuses on different components of posture and gait [9]. This rating scale is among the best predictors of falls [6]. A drawback is that composite scores are ‘post-hoc’ mathematical creations. The individual tests do not challenge multiple components at the same time, but remain focused on separate elements. Truly simultaneous assessment of multiple postural components has been described in experimental settings. Many groups investigated the influence of a secondary task on gait and balance in elderly subjects [10–13] or patients with balance disorders [14–21]. Most investigators focused on gait or balance control during a secondary cognitive task, although some used a secondary motor task [11,15,16,22,23]. A general disadvantage was the focus on balance control in a laboratory setting, rather than a clinically relevant environment. Although no two studies are comparable, the weight of the evidence shows that balance and gait deteriorate when a secondary task is performed. This suggests that even highly practised and seemingly ‘automatic’ processes such as walking require some degree of cognitive processing. The more complex and more novel the postural task, the higher the attentional demands [12,24]. These studies also suggested that ‘dual task’ or truly multiple task performance [25] should be used to predict falls. Lundin-Olsson et al. [26] studied this first. They reported that elderly subjects, who stopped walking when talking had an increased risk of falling. In fact, ‘stops walking when talking’ emerged as one of the best predictors of falls identified so far, particularly for subjects with cognitive impairment. A clear advantage
is the easy applicability by clinicians. A drawback is that the test seems unremarkable in persons without cognitive impairment and may not predict falls caused largely by motor disability (as in Parkinson’s disease) [27]. This may restrict wider practical use. The investigators later showed that difficulty with a secondary manual task (carrying a glass of water) also predicted falls in the elderly [11]. However, most subjects were demented or depressed, leaving unanswered whether falls in cognitively intact persons can be predicted by combining two motor tasks. We wanted to examine the prediction [25] that tests probing multiple (i.e. more than two) postural components would be more sensitive than a strictly dual task design. Interestingly, complex secondary tasks may distinguish better between balance-impaired patients than simpler secondary tasks [15]. Indeed, strictly dual task designs do not always distinguish well between patients and controls, over and above any baseline differences between these groups [27,28]. We speculated that combinations of various motor tasks would be particularly useful for patients without cognitive impairment, because their falls are not predicted by combining a single motor task with a mental task [27]. We further reasoned that falls in daily life would be predicted best by tests that represent complex everyday situations [29]. We also argued that falls would be predicted best by tests that truly challenge postural safety. Finally, we wanted to develop a balance test that would potentially be easy to apply in a consulting room by clinicians. Therefore, our first goal (part A of this paper) was to describe the development of a new balance test that fulfilled the above criteria. Our second goal (part B) was to evaluate normal coping strategies with increasingly complex postural tasks. One possibility is that impaired multiple task performance reflects a limited processing capacity of the central nervous system. If this were correct, healthy subjects should be able to integrate fairly complex postural tasks without errors (although errors would inevitably appear for extremely complex tasks). Alternatively, during complex tasks, healthy subjects might purposely lend priority to complete certain task components at the expense of others [30]. According to this view, a blockade would not reflect pathology but ‘prudent’ behaviour intended to optimise the primary task (maintaining balance). This strategy is termed ‘posture first’ [12]. If such intended ‘priority processes’ exist, subjects should portray slowness or a block in executing certain components of complex postural tasks. Theoretically, the safest postural strategy would be to favour maintaining balance (the ‘primary’ task) over execution of e.g. a manual or mental task. Better insight in these normal strategies is a prerequisite for interpreting pathological processes in balance disorders. To study aging effects, we included both young and elderly subjects.
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
2. Part A: development of a new ‘multiple tasks’ test
2.1. Methods We first identified relevant risk factors for falls from an orienting literature review. Secondly, we identified actual fall circumstances from an earlier prospective survey in Parkinson’s disease [31,32]. These risk factors and fall circumstances were then ‘transformed’ into functional tests (or postural ‘components’) that resembled everyday situations. These components were subsequently combined to yield the MTT.
2.1.1. Literature re6iew We performed an orienting literature review to identify relevant risk factors that could be used to design functional balance tasks. For this purpose, we applied several restrictions to our review. First, retrospective studies were excluded because elderly subjects easily forget falls [33]. Second, we only included analyses of multiple (] 2) falls, injurious falls, or both. Multiple falls are a better index of chronic disorders than single falls, which are often caused by environmental accidents with a low recurrence rate [34] and have little clinical importance, unless injury occurs. Third, because risk factors for falls are often interrelated, we confined the review to risk factors that were independently (in multivariate analyses) associated with falls. Fourth, we restricted our review to risk factors that were consistently (across studies) associated with falls in the elderly. Finally, risk factors were only included if they could be transformed to functional tasks for use in a consulting room. Since our review was orienting, we did not take the methodological quality of the selected papers into account. 2.1.2. Fall circumstances These were obtained from a separate prospective study on falls in 59 Parkinson patients (mean age 61 years; 21 women; mean Hoehn and Yahr score 2.3) and 55 healthy controls (mean age 60 years; 37 women) [31,32]. Subjects recorded the exact circumstances of all falls during 6 months, using standardised scoring forms that were returned directly after each fall. Subjects were also contacted by telephone every 2 weeks to assure that no falls were missed. 2.2. Results 2.2.1. Risk factors for falls Table 1 shows the risk factors independently associated with repeated or injurious falls in the elderly. Several risk factors were patient-related (e.g. visual impairment), whereas others were environmental factors (e.g. poor illumination). Many other risk factors were inconsistently associated with falls. Some risk
193
factors were excluded because they could not be transformed into functional tests (e.g. use of psychotropic medication) or could not be used in consulting rooms (e.g. climbing stairs). We also excluded risk factors that produced tests which were difficult to standardise or score (e.g. ‘abnormal reaction to push or pressure’) [35].
2.2.2. Fall circumstances Patients reported 205 falls, and controls reported 10 falls. Reliable information about fall circumstances was available for 160 falls of patients and all falls of controls (Table 2). 2.2.3. The multiple tasks test (MTT) 2.2.3.1. Task components. Based upon the literature review and fall circumstances, we selected 11 ‘components’ that could be used to design functional tasks for use in a consulting room (Table 3). Several interrelated risk factors or circumstances were combined to yield a single component. For example, we combined the two risk factors ‘poor illumination’ and ‘visual impairment’ into a component where light in the examination room was reduced. Similarly, we combined ‘dizziness upon Table 1 Risk factors identified in the orienting literature review Identified risk factors
Functional components
Poor illuminationa Visual impairmentb Domestic environmenta Dizziness upon standingb Orthostatic hypotensionb Problems rising from a chairb Leg weaknessb
Reduced illumination of test environment Reduced illumination of test environment Living room sitting; obstacles on floor Standing up from a chair Standing up from a chair Standing up from a chair
Standing up from a chair; squatting and touching the floor Lower extremity Standing up from a chair; squatting and disabilityb touching the floor Gait impairmentb Undisturbed walking; avoiding obstacles Stops walking when Performance of a simultaneous mental task talkingb Cognitive Performance of a simultaneous mental task impairment/demen tiab Turningb Turning 180° Presence of multiple Combinations of the above tests risk factorsa,b The second column shows the functional ‘components’, into which these risk factors could be transformed. Examples of excluded factors included (psychotropic) medication, climbing stairs, dressing impairment, reaching, decreased neck range of motion and abnormal reaction to push or pressure. a Environmental factors. b Patient-related factors.
194
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Table 2 Fall circumstances in moderately affected patients with Parkinson’s disease Fall circumstances
Functional components
Walkinga Obstacles on the floorsb Wearing shoes without proper gripb Slippery floorb Standing upa Carrying objects in the handsa,b
Undisturbed walking Avoiding obstacles Shoes with a slippery sole
Squatting, bending downb While turning arounda Doing at least two things simultaneouslya,c
Shoes with a slippery sole Standing up from a chair Carrying an empty or loaded tray Squatting and touching the floor Turning around 180° Combinations of the above components
The second column shows the functional ‘components’, into which these fall circumstances could be transformed. Examples of excluded circumstances included reaching, climbing stairs and freezing. a Patient-related factors. b Environmental factors. c This occurred in 79 out of 160 falls (49.4%) in Parkinson patients.
standing’, ‘orthostasis’ and ‘inability to rise from a chair’ into a component where subjects were asked to stand up from a chair. Different types of components could be distinguished. The first type was a cognitive component that consisted of a continuous mental task. The second type consisted of components that (largely) challenged the Table 3 Components selected for use in the Multiple Tasks Test are shown in the first column, while the respective tasks are shown in the top row
The table also shows which components were used (indicated by a ‘+’ sign) during each of the eight consecutive tasks. The shaded areas index the components that were used for scoring purposes.
motor system. This included standing up from a chair, undisturbed walking, carrying an empty tray, squatting and touching the floor, turning around and sitting down on a chair. Some components also demanded particular attention, e.g. carrying a loaded tray and avoiding obstacles on the floor. Carrying a loaded tray resembled the tasks (carrying a glass of water or a tray with glasses) used by others [11,15]. The third type was a visual component (reduced illumination in the room). The final component consisted of wearing shoes with slippery soles. These slippery shoes were promising because confrontation with new footwear may unveil multiple task difficulties that would otherwise remain unnoticed due to compensatory mechanisms, particularly in patients with longstanding disease and gradually developing lesions [18]. It was difficult to choose a proper cognitive task that continuously challenged mental processes. During pilot studies, we used open questions (e.g. ‘name as many trees as possible’). However, such questions had a disproportionate influence on task performance, as even young controls frequently blocked all movements (including answering). Conversely, asking subjects to produce serial numbers (e.g. 3-6-9-12) proved too simple and was seemingly processed automatically. Others also found that relatively simple arithmetic tasks produced little interference with concurrent, competing tasks [20]. Counting backward (e.g. serial sevens, starting from 100) produced considerable inter-individual differences. We, therefore, chose for a continuous series of relatively simple questions regarding everyday situations (e.g. ‘What is the date?’ or ‘What did you have for breakfast this morning?’). The examiner walking besides the subject posed each next question (from a standard list of 150 different questions) directly after the answer to the earlier question was given. This provided a continuous mental challenge that produced comparable results among a homogenous group of young controls. Given the difficulties in selecting an appropriate secondary cognitive task, we did not attempt to include a second, different cognitive task. Note also in this respect that Haggard et al. [20] recently failed to find relevant differences between four very different secondary cognitive tasks.
2.2.3.2. Functional tasks. The 11 components were combined to yield eight sequential tasks of increasing difficulty, due to simultaneous challenge of an increasing number of components. The MTT was performed in a quiet room (8× 3 m, linoleum floor) that was adequately illuminated. A chair was placed at each end of the room. Three obstacles (two were 9 cm wide and 3 cm high, one was 36 cm wide and 1.5 cm high) were positioned on the floor at variable distances (between 1 and 2 m). Performance was recorded on videotape.
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Undisturbed walking and ‘stops walking when talking’, which had been tested during baseline examination, were uninformative in our subjects. We, therefore, made the first task slightly more complicated by asking subjects to stand up from a chair, walk undisturbed along a predefined course, turn 180° and sit down again. This task was repeated seven times, but each time an extra component was added to the earlier and otherwise identical task (Table 3). During the second task, subjects answered a continuous series of brief questions while walking. During the third task, subjects avoided the obstacles on the floor. During the fourth task, subjects carried an empty tray. During the fifth task, the tray was loaded with two hardboiled eggs in cups and one loosely rolling egg. During the sixth task, subjects wore indoor shoes with slippery soles. During the seventh task, subjects squatted and tipped the floor halfway the obstacle course. During the eighth task, subjects wore sunglasses, while illumination was moderately reduced. Use of sunglasses allowed us to leave the room sufficiently lit for videotape recording. Subjects were instructed not to prioritise any given component, but to combine all components of each task as good as possible, at their own preferred speed. Most motor components could be executed simultaneously, although some were in fact executed directly after each other, such as touching the floor while walking, or sitting down after walking. Unlike some other studies [22], we urged subjects only once (at the beginning of the experiment) to not purposely prioritise any given component. If this instruction is continuously repeated, one might theoretically obscure any tendency to ‘disobey’ the initial instruction and to lend priority to what subjects perceive as the primary task (e.g. maintaining balance). Study of such priority strategies was a main goal of our study. During all tasks, the investigator walked beside the subject to prevent falls. We considered using a safety harness attached to a low-friction overhead track, but rejected this because it would hamper practical use in a consulting room.
2.2.3.3. Scoring system. Scoring was partially based upon subjective (qualitative) interpretation of subjects’ performance on separate task components. The simplicity of this approach has clear advantages for use in a clinical setting. Moreover, quantitative scoring during the test (e.g. with a stopwatch) would be unpractical because many test components had to be executed (and scored) continuously. Four components (carrying the unloaded or loaded tray, wearing slippery shoes and reduced illumination) could not be scored independently, but served to complicate the task and thus facilitate production of Errors. For scoring purposes, the other components were divided
195
into ‘motor’ (standing up, sitting down, walking, avoiding obstacles, turning around, touching the floor) and ‘cognitive’ components (answering questions). Impaired multiple task performance can be reflected by slowing [11,17,23] or a complete stop [26,36] in executing one or more components. Therefore, all tasks were scored as follows: rapid performance of all components within the task (‘Normal’); obvious slowing in one or more components within the task (‘Hesitation’); complete stop or inability to perform one or more components within the task (‘Block’). For example, a Hesitation was scored if subjects clearly answered the questions slower than their own baseline answering performance, which was first determined for each individual while seated. This baseline performance always produced direct answers to the simple questions without any hesitation in seated subjects, which were screened to exclude disorders that might affect the ability to answer, such as impaired cognition or dysarthria. Compared with this baseline performance, hesitations (delays) in answering were always very obvious during the multitasking conditions (silence for at least 1 s, or ‘uh-uh’ sounds). A Block was scored if subjects entirely stopped answering questions. Similarly, performance during walking, standing up, turning and sitting down was compared with baseline performance during the Tinetti Mobility Index. Hesitations and Blocks will be analysed separately, but will also jointly be referred to as ‘Errors’. The score was determined for all eight consecutive tasks of the MTT. Since we were interested in individual performance, our scoring system produced the proportion of subjects with either a completely error-free performance, as well as the proportion of subjects that made at least one Error during any given test. Hence, subjects received an abnormal test score if they made at least one Error (Hesitation or Block) during a given task. Conversely, subjects only received a normal score if they performed all components within a given task without any Error. Absolute numbers of Errors (Hesitations or Bocks) for each task were not scored because scoring individual performance is more helpful from a clinical perspective as a diagnostic tool. Scoring was done directly during the tasks. In addition to the above-described qualitative scoring, we also quantified movement time objectively using a stopwatch. For this purpose, performance was recorded on videotape. Although subjects were left free to execute the tasks at their own preferred speed, we used these videotapes to quantify the time between start (standing up) and end of each task (seated position) as an extra outcome variable.
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
196
Table 4 Performances for motor components within each of the eight MTT tasks Task
Young controls (N= 50) N
One Two Three Four Five Six Seven Eight
50 46 46 50 47 41 47 47
H (100) (92) (92) (100) (94) (82) (94) (94)
0 2 2 0 3 9 3 3
B (0) (4) (4) (0) (6) (18) (6) (6)
P-value ( 2)
Elderly controls (N = 13)
0 2 2 0 0 0 0 0
(0) (4) (4) (0) (0) (0) (0) (0)
N
H
13 (100) 12 (92) 12 (92) 11 (84) 10 (77) 12 (92) 9 (70) 13(100)
0 0 0 1 1 1 2 0
B (0) (0) (0) (8) (8) (8) (15) (0)
0 1 1 1 2 0 2 0
(0) (8) (8) (8) (15) (0) (15) (0)
0.67 0.67 0.02 0.02 0.34 B0.01 0.49
The numbers of subjects are shown (percentage between parentheses) with a normal, rapid performance (N), a motor Hesitation (H) or a motor Block (B). Hesitations or Blocks in the cognitive component (answering serial questions) were ignored for this analysis. The P-values refer to differences in performance between the two groups.
3. Part B: normal coping strategies for complex postural tasks
3.1. Subjects and methods Experiment 1. Fifty young subjects (29 women, mean ( 9 S.D.) age 27.69 6.6 years) and 13 elderly subjects (six women, mean age 62.097.8 years) received the MTT in order of increasing difficulty. Experiment 2. Twenty different healthy young subjects (eight women, mean age 20.192.2 years) received the MTT in order of decreasing difficulty. We used fairly strict inclusion criteria to select a rather homogeneous sample of elderly persons without any obvious physical or cognitive problems. Without strict inclusion criteria, elderly persons with underlying, partially subclinical age-related diseases might be included as well, and this could increase the variability in the data [37– 39]. Thus, for both experiments, history taking and detailed physical examination (including the Tinetti Mobility Index [9]) were used to exclude balance problems and neurological, orthopaedic, speech or visual disorders in all subjects. The Mini Mental State Examination [40] was used to exclude cognitive problems in the elderly subjects (mean score was 29.0; range 25– 30). All subjects gave informed consent as approved by the Ethical Committee of the Leiden University Medical Centre.
3.1.1. Statistical analyses A two-way (group by task complexity) MANOVA for repeated measures was used to compare the number of subjects who produced Errors (i.e. Hesitations or Blocks) for each task across young and elderly subjects. This was done separately for motor Errors and for cognitive Errors. In addition, to evaluate if cognitive Errors changed differently than motor Errors with task complexity, a three-way (error type by group by task complexity) MANOVA for repeated measures was
used. Greenhouse–Geisser Epsilon was used to correct for non-sphericity. These analyses were used after we ascertained that identical results were obtained when the data were fitted using a random-effects Poisson regression model. In addition, we compared the proportions of subjects who made Errors for each individual task using the Chi-square test ( 2-test). A similar analysis using a two-way (sequence direction by task complexity) MANOVA for repeated measures was used to compare the number of subjects who produced Errors for each task across subjects who received the MTT in order of increasing difficulty and subjects who received the MTT in reverse order. The log-rank test was used to study whether the number of subjects that performed all eight tasks without Errors differed between young and elderly subjects, and between subjects who received the MTT in order of increasing difficulty versus subjects who received the MTT in reverse order. Relative risks (and 95% confidence intervals) of making an Error in at least one component of the test were calculated using a Cox-proportional hazards model. Finally, the time taken to complete each task was compared between young and elderly subjects using a two-way (group by task complexity) MANOVA for repeated measures, followed by post-hoc comparisons using Tukey’s test to correct for multiple comparisons.
3.2. Results 3.2.1. The MTT in young and elderly healthy subjects Table 4 shows the performance of motor components. All subjects completed the MTT without falling. Four young subjects made Errors during the second task. Two of them had a Block (stopped walking). Most motor Hesitations occurred for the sixth task (when subjects wore slippery shoes for the first time). Overall, the proportion of subjects that produced Hesitations or Blocks for the motor components differed between young and elderly subjects (F(1,7)= 5.12, PB
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
0.05), more elderly subjects making Errors than young subjects. Table 4 shows that the proportion of subjects with motor Hesitations or Blocks was higher among elderly controls during the fourth, fifth and particularly the seventh task. Furthermore, the proportion of subjects that made motor Errors increased significantly as the tasks became more complex (F(1,7) = 2.66, PB 0.05). This increase differed across the two groups (significant interaction of Task by Group; F(1,7)= 3.07, P =0.01) because particularly elderly controls produced motor Errors during the more complex tasks. In all 62% of subjects in both groups performed all eight tasks without any motor Error (Fig. 1). Note that Fig. 1 provides complementary information to Table 4, which shows performance for all subjects for each task. In contrast, the survival analysis presented in Fig. 1 implies that anyone who produced an Error during a given task did not proceed to the next task. Different results were obtained for the cognitive Hesitations or Blocks. With task complexity, the proportion of subjects making cognitive Errors increased even more than for motor Errors, and this increase was now most pronounced in young subjects (significant interaction of Group by Error Type by Task; F(1,1,7) =3.85,
197
P= 0.001). Fig. 2 shows the performance when Hesitations or Blocks were scored for both the cognitive and motor components. Only 16.0% of young controls completed all eight tasks without any Error, due to an increasing number of different subjects with cognitive Errors as task complexity increased (note again the difference between this survival analysis and the group results of Table 4). The strategy of young subjects apparently was to postpone answering until the motor components had been completed safely. The elderly performed somewhat better than young subjects because more elderly subjects (30.8%) performed all eight tasks without any motor or cognitive Errors (no significant difference).
3.2.2. Mo6ement time Movement time increased significantly as the tasks became more complex (F(1,7)= 187.79, PB0.001). This increase in movement time differed across both groups (significant interaction of Group by Task; F(1,7)= 5.02, P= 0.001) because movement time increased more steeply in elderly subjects than young subjects. The total time to complete the MTT differed between both groups (F(1,7)= 10.79, PB 0.005) be-
Fig. 1. Kaplan – Meier curves for the cumulative proportion of subjects with a completely Error-free performance for all motor components within each respective task of the MTT. Subjects who made an Error (Hesitation or Block) for at least one motor component of any given task were excluded from the following tasks. Errors in the cognitive component (answering serial questions) were ignored for this analysis. In both groups, 62.0% of subjects had an Error-free performance.
198
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Fig. 2. Kaplan – Meier curves for the cumulative proportion of subjects with a completely Error-free performance for all components (both motor and cognitive) within each respective task of the MTT. Subjects who made an Error for at least one component of any given task were excluded from the following tasks. Sixteen percent of the young controls and 30.8% of the elderly controls completed the MTT without any Errors (no significant difference).
cause elderly subjects performed all eight tasks slower than young controls. Particularly the time to complete the eighth and most difficult task was increased in elderly subjects (mean 23.8 s, range 15.7– 34.1) compared with young subjects (mean 18.5 s, range 10.8– 26.2; P B0.05).
3.2.3. The MTT in re6erse order Motor Errors were more common among subjects who received the MTT in reverse order than among subjects who received the MTT in order of increasing difficulty (F(1,7)= 5.90, P B0.05). The proportion of subjects making motor Errors increased as the tasks became more complex in both groups (F(1,7) =3.25, P B 0.01), and this increase differed across the two groups (significant interaction of Sequence by Task; F(1,7)= 2.96, P =0.01). Particularly the two most difficult tasks produced more subjects making motor Errors among those who directly received them at the beginning of the experiment (Table 5). Conversely, virtually no subjects made Errors during the two simplest tasks in both groups. This is differently illustrated (survival analysis) in Fig. 3, which shows that only 40% of subjects made no motor Errors for the reverse MTT,
as opposed to 62% for those who received the tasks in order of increasing difficulty. Compared with subjects who received the tasks in order of increasing difficulty, the relative risk of making an Error in at least one motor component of the reverse MTT was 2.45 (95% confidence interval 1.18–5.08; PB 0.05).
4. Discussion
4.1. Strategies in healthy subjects This study shows that healthy subjects manifest Hesitations or Blocks while performing complex tasks. This occurred not only in elderly subjects, but also in young subjects. Since all subjects had a normal balance, this suggests that a normal strategy is to lend priority to complete certain aspects of a complex task, at the expense of others. As task complexity increased, more subjects made motor and cognitive Errors. Complete Blocks occurred less often than Hesitations, and no subject fell. These observations suggest that Errors in multitasking are not necessarily a marker of postural instability or pathologically impaired central processing
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
capacity. Instead, in healthy subjects, Hesitations or Blocks made during a complex task may reflect ‘prudent’ behaviour, designed to optimise postural control and to avoid falls.
4.2. Influence of ageing The MTT revealed some differences between young and elderly subjects. More elderly subjects made motor Errors, in particular during some of the most complex tasks. In contrast, fewer elderly subjects seemed to make cognitive Errors during the MTT. This suggests that young subjects were perhaps more inclined to use a ‘posture first’ strategy than elderly subjects. Others also noted that particularly young subjects favour execution of the postural task at the expense of cognitive errors [22,24,29]. The ‘posture first’ strategy is chosen particularly if the postural task is perceived as hazardous [12,24], as likely occurred during the MTT. Similar priorities are made under many everyday circumstances, e.g. car drivers who cease talking while approaching a hazardous crossing [30]. We also found that elderly subjects performed all tasks somewhat slower than younger subjects. Young subjects possibly paid the price for their ‘speedy’ performance by making more cognitive Errors during complex tasks than elderly subjects. Overall, the differences between young and old subjects were relatively subtle. One explanation is that our elderly subjects were younger (mean age 62 years) than those studied by e.g. Lundin-Olsson et al. [26] (mean age 80 years), Woollacott et al. [41] (mean age 79 years), Brown et al. [42] (mean age 79 years) and Shumway-Cook et al. [12] (mean age 74 years). Another reason is that we studied relatively few elderly subjects, certainly compared with the much larger group of young controls. Variability in performance increases
199
with aging, and larger sample sizes might provide better insight into the range of this variability and perhaps their underlying causes. However, there is mounting evidence that this variability is not primarily caused by aging itself, but rather by underlying (and perhaps otherwise subclinical) diseases [37–39]. Variability is considerably less in carefully selected elderly subjects [43]. We tried to study aging itself by including a rather homogeneous sample of carefully selected elderly persons without obvious physical or cognitive problems upon clinical examination. Indeed, none of our elderly subjects had balance impairment or cognitive deterioration, and performance was rather consistent across elderly persons. This careful selection probably also explains the modest differences with the younger subjects. We do not believe that increasing the number of elderly controls would have led to fundamentally different insights. However, it will be important to perform future studies in less healthy old people to further unravel the contributions of aging and disease to multiple task performance. Impaired multiple task performance likely reveals more abnormalities in patients with balance disorders or cognitive deficits. Indeed, our first experience indicates that the MTT can clearly distinguish patients with Parkinson’s disease from healthy elderly controls [44].
4.3. Learning effects It is possible that motor learning (performance gain through practice) influenced the first experiment, where subjects consistently received tasks identical to earlier ones, except for one novel component. We, therefore, performed a second experiment, where young subjects received the MTT in reverse order (i.e. the most difficult task first, while successive components were eliminated for each of the next tasks). We reasoned that
Table 5 Performances in subjects who received the MTT in order of increasing difficulty versus subjects who received the eight tasks in order of decreasing difficulty Task
Increasing difficulty (N= 50) N
Motor errors One Two Three Four Five Six Seven Eight
50 46 46 50 47 41 47 47
E
(100) (92) (92) (100) (94) (82) (94) (94)
P-value ( 2)
Decreasing difficulty (N =20)
0 4 4 0 3 9 3 3
N
(0) (8) (8) (0) (6) (18) (6) (6)
20 20 18 18 16 17 14 14
E
(100) (100) (90) (90) (80) (85) (70) (70)
0 0 2 2 4 3 6 6
(0) (0) (10) (10) (20) (15) (30) (30)
– 0.43 0.43 0.08 0.06 0.21 0.02 0.01
Subjects with a Hesitation (H) or a Block (B) are shown together as Errors (E). The remaining subjects had a normal, rapid performance (N). Numbers of subjects (percentage between parentheses) are shown. Only Errors for the motor components are shown in this table, but a similar pattern emerged when both motor and cognitive Errors were scored. The P-values refer to differences in performance between the two groups.
200
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
Fig. 3. Comparison between 20 young subjects who received the MTT in order of decreasing difficulty (most difficult task first), as opposed to the 50 young subjects described earlier who received the MTT in order of increasing difficulty (simplest task first).
if learning effects were present, the most difficult task would produce more Errors in subjects who directly received this task without carry-over from earlier, less complex tasks. This was indeed observed. Subjects who received the most difficult tasks first made more Errors than subjects who received these tasks at the end. We, therefore, conclude that learning explained at least part of the results during the first experiment, at least in the younger subjects.
4.4. The Multiple Tasks Test: methodological aspects As reviewed in Section 1, others also investigated the influence of secondary tasks on gait and balance in elderly subjects and patients with a variety of diseases. Similar to our findings, these studies showed that dual task performance negatively affects gait and balance. However, our study is one of the first to examine the effect of multiple (more than two) tasks on balance and gait. An advantage of our approach is that the MTT is based upon complex situations that subjects may encounter in daily life. One might argue that the most complex tasks no longer resembled everyday situations. However, our analysis of fall circumstances showed that these tasks were not as far-fetched as they may initially seem [32]. Thus, falls in Parkinson patients typically occurred under fairly complex situations (e.g. carrying a loaded tray into dimly lit rooms with doorsteps and other obstacles, wearing inappropriate
footwear and while talking to a partner). An advantage of the difficulties experienced by healthy subjects is that the MTT produces measurable results even in control groups. This may prove beneficial for its use as a diagnostic tool. Geurts et al. [29] suggested that a multiple task design should contain the following components: perceptual manipulations (e.g. distorted visual information), cogniti6e manipulations, motor manipulations (e.g. turning) and mechanical manipulations (e.g. avoiding obstacles). These manipulations should be combined to produce complex environmental conditions. The MTT fulfilled all these requirements. In contrast to others [11], we did not emphasise speed of performance, again because we wanted to replicate daily life situations. Under normal circumstances, elderly subjects are likely to trade off velocity for safety and adopt a slower and more secure performance [45]. Yet, movement time can still be measured as an extra outcome variable, and this indeed distinguished young from elderly subjects. An important goal was to study concepts underlying multiple task performance, and the current MTT served that purpose. Of course, in its present form the MTT is not an ‘end-product’ and has shortcomings. One drawback is the subjective scoring system. While advantageous for clinical use in a consulting room (no complicated equipment is necessary), it is potentially subject to individual bias. Using the videotapes, we are
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202
now investigating intra- and inter-observer reliability. Another drawback is the duration of the complete MTT, although even elderly subjects required only a few minutes to complete it. Furthermore, the sequence in which the respective components were added to the tasks was only one of many possibilities. However, administering the separate components in a random sequence would have required a much larger sample size to obtain statistically meaningful results for each test sequence. We hope to simplify the MTT, e.g. by evaluating which components are most informative in balance impaired patients. Perhaps balance problems might be identified earlier when subjects first receive more complex tasks. This would obviate the necessity for less complex tasks and shorten the test considerably.
Acknowledgements G. van der Giessen is gratefully acknowledged for his expert assistance. We thank Dr J.G. van Dijk and Dr D.J. Beckley for their critical comments, and Dr A.H. Zwinderman for his statistical analyses.
References [1] Maki BE, Holliday PJ, Topper AK. Fear of falling and postural performance in the elderly. J Gerontol 1991;45:M123 –31. [2] Bloem BR, van Vugt JPP, Beckley DJ. Postural instability and falls in Parkinson’s disease, Adv Neurol 2001;87:209 –23. [3] Sattin RW. Falls among older persons: a public health perspective. Annu Rev Publ Health 1992;13:489 – 508. [4] Banks MA, Caird FI. Physiotherapy benefits patients with Parkinson’s disease. Clin Rehab 1989;3:11 –6. [5] Tinetti ME, Baker DI, McAvay G, et al. A multifactorial intervention to reduce the risk of falling among elderly people living in the community. New Engl J Med 1994;331:821 – 7. [6] Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. New Engl J Med 1988;319:1701 – 7. [7] Tromp AM, Smit JH, Deeg DJH, et al. Predictors for falls and fractures in the longitudinal aging study Amsterdam. J Bone Miner Res 1998;13:1932 –9. [8] Horak FB, Shupert CL, Mirka A. Components of postural dyscontrol in the elderly: a review. Neurobiol Aging 1989;10:727 – 38. [9] Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc 1986;34:119 –26. [10] Maylor EA, Wing AM. Age differences in postural stability are increased by additional cognitive demands. J Gerontol Psychol Sci 1996;51B:P143 – 54. [11] Lundin-Olsson L, Nyberg L, Gustafson Y. Attention, frailty, and falls: the effect of a manual task on basic mobility. J Am Geriatr Soc 1998;46:758 –61. [12] Shumway-Cook A, Woollacott MH, Kerns KA, Baldwin M. The effects of two types of cognitive tasks on postural stability in older adults with and without a history of falls. J Gerontol Med Sci 1997;52A:M232 –40.
201
[13] Marsh AP, Geel SE. The effect of age on the attentional demands of postural control. Gait Posture 2000;12:105 –13. [14] Camicioli RM, Oken BS, Sexton G, et al. Verbal fluency task affects gait in Parkinson’s disease with motor freezing. J Geriatr Psychiatry Neurol 1998;11:181 – 5. [15] Bond JM, Morris M. Goal-directed secondary motor tasks: their effects on gait in subjects with Parkinson disease. Arch Phys Med Rehab 2000;81:110 – 6. [16] Alexander NB, Mollo JM, Giordani B, et al. Maintenance of balance, gait patterns, and obstacle clearance in Alzheimer’s disease. Neurology 1995;45:908 – 14. [17] Camicioli RM, Howieson DB, Lehman S, Kaye J. Talking while walking. The effect of a dual task in aging and Alzheimer’s disease. Neurology 1997;48:955 – 8. [18] Geurts AC, Mulder TW, Nienhuis B, Rijken RA. Influence of orthopaedic footwear on postural control in patients with hereditary motor and sensory neuropathy. Arch Phys Med Rehab 1992;73:569 – 72. [19] Courtemanche R, Teasdale N, Boucher P, et al. Gait problems in diabetic neuropathic patients. Arch Phys Med Rehab 1996;77:849 – 55. [20] Haggard P, Cockburn J, Cock J, et al. Interference between gait and cognitive tasks in a rehabilitating neurological population. J Neurol Neurosurg Psychiatry 2000;69:479 – 86. [21] Andersson G, Yardley L, Luxon L. A dual-task study of interference between mental activity and control of balance. Am J Otol 1998;19:632 – 7. [22] Chen HC, Schultz AB, Ashton-Miller JA, et al. Stepping over obstacles: dividing attention impairs performance of old more than young adults. J Gerontol Med Sci 1996;51:M116 – 22. [23] Means KM, Rodell DE, O’Sullivan PS. Obstacle course performance and risk of falling in community-dwelling elderly persons. Arch Phys Med Rehab 1998;79:1570 – 6. [24] Lajoie Y, Teasdale N, Bard C, Fleury M. Attentional demands for static and dynamic equilibrium. Exp Brain Res 1993;97:139 – 44. [25] Mulder TW, Berndt H, Pauwels J, Nienhuis B. Sensorimotor adaptability in the elderly and disabled. In: Stelmach GE, Ho¨ mberg V, editors. Sensorimotor Impairment in the Elderly. Amsterdam: Kluwer, 1993:413 – 26. [26] Lundin-Olsson L, Nyberg L, Gustafson Y. Stops walking when talking as a predictor of falls in elderly people. Lancet 1997;349:617. [27] Bloem BR, Grimbergen YAM, Cramer M, Valkenburg VV. Stops walking when talking does not predict falls in Parkinson’s disease. Ann Neurol 2000;48:268. [28] Morris ME, Iansek R, Smithson F, Huxham F. Postural instability in Parkinson’s disease: a comparison with and without a concurrent task. Gait Posture 2000;12:205 – 16. [29] Geurts AC, Mulder TW, Nienhuis B. From the analysis of movements to the analysis of skills. J Rehab Sci 1991;4:9 –12. [30] Mulder TW, Geurts AC. The assessment of motor dysfunctions: preliminaries to a disability-oriented approach. Human Movement Sci 1991;10:565 – 74. [31] Bloem BR, Grimbergen YAM, Cramer M, Willemsen M, Zwinderman AH. Prospective assessment of falls in Parkinson’s disease. J Neurol 2001; in press. [32] Willemsen MD, Grimbergen YAM, Slabbekoorn M, Bloem BR. Vallen bij de ziekte van Parkinson: vaker door houdingsinstabiliteit dan door omgevingsfactoren. Ned Tijdschr Geneeskd 2000;144:2309 – 14. [33] Cummings SR, Nevitt MC, Kidd S. Forgetting falls. The limited accuracy of recall of falls in the elderly. J Am Geriatr Soc 1988;36:613 – 6. [34] Nevitt MC, Cummings SR, Kidd S, Black D. Risk factors for
202
[35]
[36]
[37] [38]
[39]
[40]
B.R. Bloem et al. / Gait and Posture 14 (2001) 191–202 recurrent nonsyncopal falls. A prospective study. J Am Med Assoc 1989;261:2663 –8. Bloem BR, Beckley DJ, van Hilten JJ, Roos RAC. Clinimetrics of postural instability in Parkinson’s disease. J Neurol 1998;245:669 – 73. Morris ME, Iansek R, Matyas TA, Summers JJ. Stride length regulation in Parkinson’s disease. Normalization strategies and underlying mechanisms. Brain 1996;119:551 –68. Rowe JW, Kahn RL. Human aging: usual and successful. Science 1987;237:143 – 9. Howieson DB, Holm LA, Kaye JA, et al. Neurologic function in the optimally healthy oldest old: neuropsychological evaluation. Neurology 1993;43:1882 –6. Bloem BR, Gussekloo J, Lagaay AM, et al. Idiopathic senile gait disorders are signs of subclinical disease. J Am Geriatr Soc 2000;48:1098 – 101. Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A
[41]
[42]
[43] [44]
[45]
practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189 – 98. Woollacott MH, Shumway-Cook A. Attentional demands in postural tasks: changes in both healthy and balance impaired adults. Gait Posture 1999;9(Suppl 1):S12. Brown LA, Shumway-Cook A, Woollacott MH. Attentional demands and postural recovery: the effects of aging. J Gerontol Med Sci 1999;54A:M165 – 71. Calne DB, Eisen A, Meneilly A. Normal aging of the nervous system. Ann Neurol 1991;29:206 – 7. Bloem BR, Valkenburg VV, Slabbekoorn M, Van Dijk JG. The Multiple Tasks test. Strategies in Parkinson’s disease. Exp Brain Res 2001;137:478 – 96. Patla AE. A framework for understanding mobility problems in the elderly. In: Craik RL, Oatis CA, editors. Gait Analysis. Theory and Application. St. Louis: Mosby Year Book, 1994:436 – 49.