| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Research Reports |
H van den Berg-Emons, PhD (Health Science), is Research Scientist, Institute of Rehabilitation Medicine, Faculty of Medicine and Health Sciences, Erasmus University Rotterdam, Dr Molewaterplein 40, 3015 GD Rotterdam, the Netherlands (vandenberg{at}revd.azr.nl).
J Bussmann, PhD (Medicine and Health Science), BSc (PT), is Research Scientist, Institute of Rehabilitation Medicine, Faculty of Medicine and Health Sciences, Erasmus University Rotterdam
A Balk, PhD (Cardiology), MD, is Cardiologist, Thoraxcenter, University Hospital Rotterdam
D Keijzer-Oster is a graduate student in medical science at Erasmus University Rotterdam
H Stam, PhD (Medicine and Health Science), MD (Medicine and Health Science), is Professor and Director, Institute of Rehabilitation Medicine, Erasmus University Rotterdam, and Department of Rehabilitation, University Hospital Rotterdam
Address all correspondence to Dr van den Berg-Emons
Submitted July 26, 2000;
Accepted March 12, 2001
 | Abstract |
|---|
=3.9, SD=1.5, range=2.2–6.7) than in the comparison subjects (=11.3, SD=3.0, range=6.6–14.1). In the patients, between-day variance was smaller for different weekdays (eg, Monday versus Tuesday) than for similar weekdays (eg, 2 Mondays) (1.11% and 7.28%, respectively). Discussion and Conclusion. The results show how activities associated with mobility during everyday life may be restricted in people with CHF.
Key Words: Accelerometry • Ambulatory monitoring • Between-day variance • Physical activities
| Introduction |
|---|
 TopAbstract Introduction MethodResultsDiscussionConclusionReferences |
|---|
reduced fitness early fatigue further hypoactivity. Measurement of everyday activities associated with mobility is important in managing people with CHF because it provides information on disability and prognosis.4 Furthermore, we believe that everyday activities are related to quality of life. Until now, only a few studies were available on everyday activities in people with CHF. Methods that have been used include the use of an actometer,2,5 a pedometer,4,6,7 a calorimeter,8 and the doubly labeled water technique.9,10 These methods, however, provide only information on the level (or intensity) of everyday physical activity. They provide no information on the activities performed. Commonly used methods for people with CHF such as exercise tolerance testing and use of the New York Heart Association functional classification11 have been found to be inadequate in predicting actual function.2,4,5
An "Activity Monitor" (AM) that provides information on several aspects of activities associated with mobility has been developed.12,13 The AM is based on more than 24 hours of ambulatory monitoring of signals from accelerometers fixed to the body. From these signals, the duration, rate, and moment of occurrence of activities associated with mobility (eg, lying, sitting, standing, walking [including walking up and down stairs], running, cycling, wheelchair use, general movement) and transitions (changes in posture) can be detected with a 1-second resolution. Information on the variability of the acceleration signal (motility) can be obtained, which is related to the intensity of body-segment movements.14–16 Apart from monitoring accelerations, other signals such as heart rate or electrocardiographic activity can also be measured by the device.
The aim of our study was to obtain information on the level of activities associated with mobility during everyday life of people with mild to moderate chronic CHF as measured with the AM. Furthermore, we examined the between-day variance in activities because we believe that this information is important in intervention studies for the determination of the optimal number of monitoring days and the required sample size. Because we expect a weekly activity pattern (eg, shopping on Mondays, housekeeping on Tuesdays, and so on), we also studied whether measurements of between-day variance in activities can be reduced by monitoring on similar weekdays (eg, on 2 Mondays rather than on a Monday and a Tuesday). Our study was conducted in preparation for a large-scale intervention study on the effects of aerobic training on daily functioning in people with CHF. The research questions were:
| Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Subjects
Five subjects with stable CHF (mean age=64 years, SD=5, range=59–72) were included in the study. All subjects with CHF were male; no female patients participating in the screening project were available at the time of the study. Clinical characteristics of the subjects with CHF are presented in Table 1. All subjects in this group had had symptoms of CHF for at least 1 year, were retired from work, and were living with a partner.
|
Device (Fig. 1)
Activities are detected using predetermined criteria written into a custom-made software program. Reliability and validity have been investigated in previous studies.13,17–19 We believe that the only sources of error that might possibly affect the test-retest reliability are changes in the attachment of the sensors to the body (exact location on body segments), instability of the sensors, and instability of the software program. We believe that errors due to changes in the attachment of the sensors were minimized because we used standard procedures for the attachment of sensors. Furthermore, extended calibrations of the sensors have revealed that the sensors are stable, even over longer periods of time. To test the stability of the software program, we performed repeated analyses of activity detection on several data files containing accelerometer signals monitored during 24-hour periods. The analyses were performed on separate days and in different periods of the year. Results of these analyses have shown that the output of the AM is identical over repeated analyses of activity detection.
|
In our study, 4 IC-3031 uniaxial piezo-resistive accelerometers* were used (size: 1 x 1 x 1 cm or 2 x 2 x 0.5 cm). One sensor was attached to each thigh, and 2 sensors were attached to the skin over the sternum. The sensors on the thighs were attached to the skin with Rolian Kushionflex
(while standing, the sensors are sensitive in an anteroposterior direction), and adhesive medical tape was used to consolidate the attachment. The sensors on the trunk were attached to the skin with silicone-based stickers
(while standing, one sensor is sensitive in an anterioposterior direction and one sensor is sensitive in a longitudinal direction). All sensors were attached as parallel as possible to the vertical or horizontal plane; a maximum deviation of 15 degrees was allowed. For a more detailed description of the sensors and the attachments, see Bussmann and colleagues12,13 and Veltink et al.20
The accelerometers were connected to a Vitaport2 data recorder (size: 15 x 9 x 4.5 cm, weight: 700 g) or a Rotterdam Activity Monitor* (RAM) (size: 15 x 9 x 3.5 cm, weight: 500 g), which were worn in a padded bag round the waist. For logistical reasons, different devices (Vitaport2 and RAM) were used; however, the most important differences between the devices were the size and weight. Accelerometer signals were stored digitally on a PCMCIA hard disk or flash card with a sampling frequency of 32 Hz. After the measurement, data were downloaded onto a Macintosh computer
for analysis. In the analysis (Signal Processing and Inferencing Language|| 3 parts could be distinguished: (1) feature processing, (2) activity detection, and (3) postprocessing.
In feature processing, 3 feature signals were derived from each measured signal. First, low-pass/angular signals were created by low-pass filtering (0.3 Hz) of the measured signals. These signals were then converted to angles (ranging from –90° to +90°). In 2 subjects, the deviation of the trunk sensor to the vertical plane was more than 15 degrees. A software program was used to correct this deviation in the angular signals. Second, a motility signal was created by high-pass filtering (0.3 Hz), rectifying, and smoothing the data. This signal depends on the variability of the measured signal around the mean (unit of motility is an arbitrary acceleration unit). Third, the frequency signal was based on a band-pass-filtered derivative (0.3–2 Hz for legs and 0.6–4 Hz for trunk) of the measured signal. This band-pass signal was the input of the Fast Time Frequency Transform (FTFT22) procedure (a type of instantaneous frequency analysis that determines the frequency of the band-pass signal). All features had a time resolution of 1 second.
Activity detection was based on the signals. Twenty-three activity subcategories were distinguished (Tab. 2). For each subcategory, a minimum value and a maximum value were preset for each signal in an activity detection knowledge base (based on studies of subjects with and without known pathology). For consecutive moments in time (1 second), for each subcategory, the distance of each feature was calculated from the actual value to the preset range. If the actual value was within the range, the distance was 0. The calculated distances were added for each activity subcategory, and the activity subcategory with the smallest distance was selected.
|
|
To interfere as little as possible with normal activity patterns, subjects were fitted at home with the AM (between 10:00 AM and 11:30 AM). During the activity monitoring, subjects were not allowed to swim or to take a bath or shower. After the measurements, we visited the subjects again to remove the instrumentation and to ask them questions about the kind of activities they had performed and the convenience of the AM. In order to avoid bias, the complete aim of the study was initially not explained to the subjects. Furthermore, subjects were instructed to continue their ordinary daily life (with the exceptions previously noted). After the measurements, complete information on the aim of the study was given to the subjects, and the reason for not giving that information before the measurements was explained. All subjects agreed with this procedure; thus, all measurements were included in the analysis.
Data Analysis
The measurements were obtained for less than 24 hours per day for 3 subjects. In 1 subject, the trunk accelerometers had come loose from the skin at the end of the measurement period (presumably due to excessive perspiration). In another subject, a problem had occurred with the batteries for the AM. In the third subject, there were incomplete acceleration signals during 1 hour. In the analysis, only corresponding measurement periods were used between days or between subjects with CHF and comparison subjects (eg, in case patient data were missing, for example between 12:00 noon and 1:00 PM on day 1, data obtained for this period on the other measurement days were excluded from the analysis and the same was done for the comparison subject). Therefore, the mean amount of time of a measurement day that was used for analysis was 19.6 hours (SD=2.0).
For logistical reasons, the first part of the 48-hour measurement in the first week corresponded to the measurement in the second week in some subjects with CHF, whereas in other subjects with CHF the second part of the 48-hour measurement corresponded to the measurement in the second week. The 24-hour measurement in the second week was called "weekday 2A" ("2" refers to week 2). The 24-hour part of the consecutive (48-hour) measurement in the first week that corresponded to this weekday was called "weekday 1A" ("1" refers to week 1). The other 24-hour period of the consecutive measurement was called "weekday 1B." Thus, weekdays 1A and 2A were similar weekdays (eg, 2 Mondays), with 1 week between measurements; weekday 1B differed from these days (eg, Tuesday). In the comparison subjects, the first 24-hour period of the consecutive measurement was called "weekday 1A" and the second 24-hour period was called "weekday 1B." To obtain information on everyday activities associated with mobility, the following variables were assessed: duration of stationary activities and duration of movement-related activities (both as a percentage of the duration of the measurement day), distribution of activities within the stationary activity category and within the movement-related activity category, total number of transitions, number of sit-to-stand transitions, mean motility during a 24-hour period (representing the level or intensity of everyday activity), mean motility during walking (representing intensity of walking, or walking speed14–16), number of walking periods, and distribution of the duration of walking periods. When comparing the subjects with CHF with the comparison subjects, the results of the 2 consecutive weekdays were used. In order to get insight into the habituation of subjects to the AM, the first 24-hour part of the consecutive measurement was compared with the second 24-hour part. Differences in the mutual distribution within the stationary activity category and within the movement-related activity category, or in the distribution of the duration of walking periods between the subjects with CHF and the comparison subjects, were tested with a multivariate analysis of variance. Other differences between groups were tested with the Mann-Whitney U test. Comparisons within the study groups were made using the Wilcoxon test.
The variable that was used for the assessment of between-day variance in activities associated with mobility was the duration that movement-related activities were performed, as a percentage of the duration of the measurement day. Information on the between-day variance was obtained with a one-way analysis of variance. In the subjects with CHF, the between-day variance for both similar and different weekdays was based on measurements obtained with 1 week between measurements: between-day variance for similar weekdays was based on weekdays 1A and 2A, and between-day variance for different weekdays was based on weekdays 1B and 2A. Differences in variance between the subjects with CHF and the comparison subjects or within the subjects with CHF (similar weekdays versus different weekdays) were tested with the F test. All statistics were done with SPSS/PC#; a probability value of P
.05 was considered to indicate a significant effect.
| Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
|
|
|
|
| Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
No female patients were available at the time of our study. We contend that, because dyspnea and fatigue are the main limiting factors in the everyday physical activities in patients with CHF, similar findings on activities associated with mobility should be expected in women with CHF. We do not have any data, however, to support that contention.
We used uniaxial accelerometers, placed on the legs and on the sternum, in our study. While the subjects were standing, the sensors were sensitive in either the anteroposterior direction or in the longitudinal direction. The feasibility study of Veltink et al20 and several validation studies13,17–19 have shown that the use of 4 uniaxial sensors in the described configuration is sufficient to detect the level of the gross daily activities (eg, walking, cycling) and postures. When using the device with some types of patients (eg, those using wheelchairs), additional sensors are placed on the lower arms. In a study by Bussmann et al,16 a strong relation was found between the variability of the accelerometer signal (motility) during walking and oxygen uptake (pooled r2=.91).
Everyday Activities Associated With Mobility
Based on the results for percentage of the time that movement-related activities were performed, number of transitions, mean motility during a 24-hour period, and number of walking periods (Tab. 3), we conclude that our subjects with CHF were considerably less active than the comparison subjects. We believe that people with CHF may decrease their physical activity to minimize the occurrence of symptoms such as dyspnea and fatigue. Furthermore, the hypoactivity observed in people with CHF may be caused by the low exercise tolerance that these individuals are known to have.23,24
From the results shown in Table 5, we conclude that there was no difference in the distribution of the duration of walking periods between the subjects with CHF and the comparison subjects. This finding is in contrast to our expectation that the subjects with CHF would prefer short-lasting walking periods as compared with the comparison subjects. The mean motility during walking, which is assumed to be related to walking speed,14–16 was not lower in the subjects with CHF than in the comparison subjects. This finding is not in line with what we expected. An explanation for the difference between the results of our study and our expectations may be that both groups of subjects spent most of the measurement time within their homes. It is likely that durations of walking periods or walking speed are then more comparable between individuals with and without CHF then when monitoring predominantly outside activities (eg, walking to shops). However, it may also be possible that our sample was too small to detect differences in these variables.
The low level of everyday activities associated with mobility that we found in our subjects with CHF has also been reported by other researchers. Toth et al10 measured free-living energy expenditure (energy expenditure during normal daily activities, not measured in the laboratory) in subjects with CHF (cachectic and noncachectic) and in comparison subjects without CHF using the doubly labeled water technique. They found that the energy expenditure for physical activity was lower in the subjects with CHF (=269 kcal/d [SD=307] in those who were cachectic and =416 kcal/d [SD=361] in those who were noncachectic) than in the comparison subjects (=728 kcal/d, SD=374). Walsh et al7 reported lower pedometer scores in subjects with CHF than in comparison subjects without CHF (=258 x 102 steps/wk [SD=45] versus 619 x 102 steps/wk [SD=67]). Davies et al5 and Hoodless et al6 also found a reduction in actometer and pedometer scores, respectively, in subjects with CHF as compared with comparison subjects without CHF.
Between-Day Variance
Information on the variance in everyday activities associated with mobility in people with CHF is important in order to assess the number of activity monitoring days that is required to get insight in the customary daily physical activity in this group. In intervention studies with paired comparisons, particularly the between-day variance is important. Based on this variance, the magnitude of the effect that the researcher wants to detect, and the available number of subjects, the required number of sampling days can be assessed. Our study was a preliminary investigation for a large-scale intervention study on effects of aerobic training on daily functioning in people with CHF. Based on the results for between-day variance obtained in this preliminary investigation and a relative increase of 33% in duration of movement-related activities that we want to detect, 2 sampling days before and 2 sampling days after the training intervention seems to be appropriate (n=35 in experimental group and n=35 in control group, power is 90%).
The between-day variance in duration of movement-related activities was relatively large in both study groups, but particularly in the comparison group (Tab. 6). In the subjects with CHF, the variance between different weekdays (with 1 week between measurements) was smaller than the variance between similar weekdays (with 1 week between measurements) (1.11% versus 7.28%, respectively). This finding is in contrast to our expectation that, on similar weekdays, similar activities would be performed (eg, shopping on Mondays [with a relatively long duration of walking], housekeeping activities on Tuesdays, and so on). Apparently, a weekly pattern of physical activity did not exist in the subjects in this study. Therefore, the assumption that monitoring during similar weekdays will reduce the between-day variance was not supported by our results. We have no explanation for the finding that the between-day variance of similar weekdays was larger than the between-day variance of different weekdays.
| Conclusion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
| Footnotes |
|---|
Ethical approval for this study was obtained from the Medical Ethics Committee of the University Hospital Rotterdam.
This study was supported, in part, by the Rotterdam Foundation for Cardiac Rehabilitation.
* Supplied by Temec Instruments BV, Spekhofstraat 2, 6460 HA Kerkrade, the Netherlands. 
Smith and Nephew Nederland, PO Box 535, 2130 AM Hoofddorp, the Netherlands.
Schwa-medico, Ehringshausen, Germany.
Apple Computer BV, Handelsweg 2, 3707 NN Zeist, the Netherlands.
|| G Mutz, Department of Psychophysiology, University of Cologne, Cologne, Germany, and WLJ Martens, Phyvision, Kromstraat 3, Gemert, the Netherlands.
# SPSS Benelux BV, PO Box 115, 2200 AC Gorinchem, the Netherlands.
| References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. K. White, R. C. Wagenaar, M. E. Del Olmo, and T. D. Ellis Test-Retest Reliability of 24 Hours of Activity Monitoring in Individuals With Parkinson's Disease in Home and Community Neurorehabil Neural Repair, July 1, 2007; 21(4): 327 - 340. [Abstract] [PDF]
|
|
|
|
 |
|
 R. J van den Berg-Emons, J. B Bussmann, A. H Balk, and H. J Stam Factors Associated With the Level of Movement-Related Everyday Activity and Quality of Life in People With Chronic Heart Failure Physical Therapy, December 1, 2005; 85(12): 1340 - 1348. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
