Vol. 10 •Issue 8 • Page 38
In Motion With Actigraphy
Standards are Needed for this Useful Tool
By Girardin Jean-Louis, PhD
Sleep professionals may be able to observe certain daytime behavior or sleep abnormalities only in the patient’s natural setting. Continuous polysomnography (PSG) in these cases can be cost-prohibitive, not to mention burdensome for the patient who must go about his day hooked up to an elaborate montage of sensors.
As an alternative, researchers and sleep technicians are using actigraphy for data collection and assessing sleep-wake patterns. Actigraphs are worn typically on the wrist or ankle to monitor body movement.
Most use analog systems, which digitize movement into activity counts and integrate them using different quantification modalities: zero crossing, time above threshold or digital integration. Some actigraphs quantify movement simultaneously in several modes. Others house additional channels to monitor ambient light levels, noise levels and body temperature.
Regardless of the modality employ ed, most recordings use an activity summation interval of one minute, and the data are stored into memory for later retrieval. The recording period varies contingent upon the study objectives and in accordance with the memory capacity of the actigraph.
To date, evidence shows that actigraphy is valid for community-residing older adults, shift-workers and individuals with insomnia or major depression.1-7 Correlation studies, without systematic optimization, have shown that actigraphy is useful for patients with sleep-disordered breathing, periodic limb movement or dementia, and to monitor infants and adolescents.7-11
Some suggest that actigraphy could detect nighttime arousals, common among patients with sleep-disordered breathing or periodic limb movement,12,13 although data from a community-based study showed that actigraphic measures correlated poorly with oxygen desaturation indices.14 Other findings offer that actigraphy is useful in monitoring response to drug therapy and other intervention approaches (e.g., sleep hygiene, afternoon naps, exercise, hot baths and light therapy).10
An actigraph can be worn unobtrusively for up to a week to record daytime physical activity itself, which may be used to differentiate between patients with attention deficit hyperactive disorder or depressives and normal controls.15 It’s used to characterize Parkinsonian tremor and to monitor sundowning behavior among patients with Alzheimer’s disease.8,10,16,17
Additionally, actigraphy can monitor the effects of pain on physical activity or the effects of different work schedules on sleep patterns of individuals in the areas of transportation, manufacturing, military and medical fields. More recently, researchers used actigraphy to monitor activity patterns of astronauts in space missions. In the laboratory or in the natural setting, actigraphy provides reliable estimates of the circadian rhythm of activity in response to experimental manipulations or in documenting environ mental influences.10
The widespread use of actigraphy has led to the recognition that several methodological issues have to be addressed to increase its acceptability, in cluding validity, reliability and sensitivity of the actigraph, placement, cross-validation, and accuracy of scoring algorithms.
To advance the discourse on the validity of actigraphy, a task force commissioned by the American Academy of Sleep Medicine (AASM) reviewed these issues and published practice guidelines in 1995 governing the use of ac tigraphy, recognizing it as a reliable adjunct to PSG in the study and clinical assessment of sleep disorders.10
Actigraphs systematically tested heretofore have yielded sleep estimates showing acceptable validity indices and agreement rates when referenced to PSG. However, while actigraphy is intended for use in the natural setting among people with sleep abnormalities, validation studies were primarily conducted among individuals without sleep disorders.
The review committee noted that validity indices and agreement rates between polysomnographic and actigraphic estimates derived from healthy sleepers were greater than those observed for clinical samples. In light of such evidence, the committee recommended more actigraphic validation studies in clinical populations.
Since this recommendation, researchers have examined the validity of actigraphy in specific clinical populations. Systematic validation studies have demonstrated improved accuracy and agreement between actigraphic and polysomnographic estimates.
Comparing Sleep Estimates
Another major methodological problem noted by the AASM review committee concerns the inherent difficulty comparing sleep estimates from various actigraphs or across laboratories. Such comparisons are hindered because actigraph manufacturers don’t follow standardized design characteristics.
As such, actigraphs differ in several respects, including movement quantification modalities, sensitivity and filter settings, motion sensors and directional sensitivities. Even within specified populations, comparisons are limited if researchers don’t use the same actigraphs.
One study comparing actigraphs from two different manufacturers found discrepancies in activity counts emanating from the same wrist, notwithstanding that motion sensors were comparable.18 A simulation study found dissimilar activity counts in response to the same activity pattern comparing activity counts derived from the three common movement quantification modalities: zero crossing, time above threshold and digital integration.19
It appears that activity counts integrated with differing modalities might not yield the same estimates, even when the same accelerometer is used. Evidence suggests that the digital integrating mode may yield the best accuracy in detecting sleep-wake states.20
It remains unclear whether tri-axial actigraphs, which have the capability of detecting subtle movement in all three directional axes, produce more accurate sleep estimates than do uni-axial actigraphs, as researchers have not performed any systematic studies. There are suggestions that tri-axial devices might represent sedentary activities more accurately, however, no difference has been found when the two were compared to indirect calorimetry.21,22
The difficulty comparing sleep estimates is compounded by the fact that there aren’t standardized criteria to score actigraphic data; each actigraph that has been tested employed a unique scoring algorithm.
To date, three scoring algorithms have been studied extensively.2-4,7,9-11,14,15,20,23
Few studies, however, have attempted to assess the accuracy of scoring algorithms when applied to actigraphic data for which they were not developed initially. These studies have shown that sleep estimates may be underestimated or overestimated, depending upon the scoring algorithm used.2,3,5,24 Sleep researchers may obtain comparable results when the same algorithm is applied to devices with comparable mode of operation.
Further, investigators have noted that scoring algorithms developed on healthy young sleepers may not necessarily yield accurate results when applied to persons with varying clinical presentations. In one study, researchers remarked that even within a specific sleep disorder classification, differing validity coefficients could be found for each subclassification.5 These findings have led to the belief that scoring algorithms must be calibrated to reflect the particularities of the population under consideration for optimal results.
Devices produced by the same manufacturer might require calibrated scoring algorithms if different transducers are used, and progress in transducer design may lead to superior actigraphs yielding more accurate results even when using old scoring algorithms. When comparing data from different actigraphs, the investigator should consider the sampling frequency, sensitivity and filter settings, motion sensor and directional sensitivity of the actigraph.
OTHER METHODOLOGICAL ISSUES
Several other methodological issues have received scientific attention. Placement of the actigraph (dominant vs. non-dominant hand) doesn’t appear to be a significant concern when estimating nocturnal sleep parameters. Data have also revealed no difference in sleep estimates comparing wrist and ankle placements.11,23
The finding that wrist and ankle activity patterns were virtually identical is important because ankle placement seems more convenient when testing infants. Ankle placement also may yield useful information in characterizing periodic limb movement disorder.10
Evidence suggests, however, that dif-ferences may be found in daytime activity recorded from the non-dominant vs. the dominant wrists, although one might suspect differences in the sensitivity of the actigraphs used. Arguably, there may be subtle differences in the sensitivity of actigraphs even when they share the same design characteristics.
Nonetheless, these don’t seem to pose a problem in differentiating sleep from wakefulness at night, particularly among individuals showing few mid-sleep awakenings. It should be noted that differences in inter-device sensitivity could be exacerbated over time when motion sensors are used extensively.
In addition, it isn’t clear how long actigraphs could be used before recalibration. One study has indicated no significant differences in sleep estimates between old (about 1.5 years) and new actigraphs.23 In all studies using more than one device, which is the current practice in research settings, it’s imperative that all actigraphs operate at the same level of accuracy and sensitivity. The actigraph also must be sturdy to detect wrist movement reliably in longitudinal designs.
Future studies should examine the confounding influences of co-sleeping and sleeping on moving surfaces on wrist-activity recordings. They also should examine the stability of actigraphic sleep measures for various clinical groups. Validation studies are encouraged for each clinical sample of interest before making broad generalizations.
The sleep field would benefit a great deal if we could establish standards for actigraph design characteristics and for scoring criteria. This would facilitate data sharing and comparisons across laboratories and across samples. Despite some of the inherent limitations of actigraphy, it’s a useful, cost-effective tool to assess sleep-related disorders and to monitor treatment responses. n
Dr. Jean-Louis is clinical director of the sleep center at Kingsbrook Jewish Medical Center and assistant professor in the departments of psychiatry and ophthalmology at SUNY Downstate Medical Center. Both facilities are located in Brooklyn, N.Y.
For a list of references please call Sharlene Sephton at (610) 278-1400, ext. 1324 or visit Respiratory-care-sleep-medicine.advanceweb.com/mrreflist.html.