Problems Using Actigraphy in People with Parkinson’s Disease
The advent of actigraphy in the 1990s made it possible to indirectly record a person’s sleep-wake cycles based on the person’s activity level, with increased activity indicating wakefulness and decreased activity indicating sleep.1,2 In actigraphy, a device — an actigraph — which is typically worn on the wrist, continually records movement data over a prolonged time — one week or more.3
Actigraphy Algorithms and Parkinson's Disease
The collected data is relayed to a computer and analyzed. It is then presented in a histogram that shows sleep and wake characteristics such as total sleep time, percent of time spent asleep, total wake time, percent of time spent awake and number of awakenings. Algorithms (i.e., specialized mathematical formulas) used in actigraphs are well-suited to measure the activity level of healthy or young people. However, the algorithms may not accurately detect sleep and wake disturbances in people with Parkinson’s disease.
Dysfunctional motor activity in Parkinson’s disease such as dyskinesia (i.e., impaired voluntary movement), bradykinesia (i.e., slow movement) and abnormal motor activity during sleep such as rapid eye movement (REM) sleep behavior disorder, can be misinterpreted by an actigraph. Some recent research indicates that using immobility (i.e., low activity) rather than activity level may more accurately assess certain aspects of sleep and wake in people with Parkinson's disease.4-6
What is Parkinson's Disease? (Signs and Symptoms)
Parkinson’s disease is a neurodegenerative brain disorder in which certain dopamine-containing cells at the base of the brain, which are involved in movement, are progressively destroyed. Why this destruction occurs is unclear.
The initial symptoms of Parkinson’s disease are muscle rigidity, akinesia (i.e., loss of voluntary movement or reduced ability to make voluntary movement), and a tremor that initially affects the fingers of one hand.
Other symptoms that occur as the disease progresses are bradykinesia, walking in a stooped position, walking with a shuffle, festinating gait (i.e., brief episode of involuntary short rapid shuffling steps), freezing (i.e., momentary inability to initiate voluntary movement), an expressionless (i.e., “mask-like”) face with a stare, difficulty in enunciation, difficulty swallowing, excess saliva production, and voice problems such as speaking in a low voice because of weakness of the muscles involved in speech.
People with Parkinson’s disease can have disrupted sleep due to degenerative brain changes or comorbid conditions, such as periodic leg movements or REM sleep behavior disorder. As a consequence, a person may struggle with excessive daytime sleepiness. The latter can also result from a drug effect (e.g., levodopa).
An actigraph contains sensors that detect the acceleration or movement of an object and converts the movement into a signal (measured in volts). An actigraph can be sensitive to vertical movements (i.e., uniaxial), sensitive to vertical and horizontal movements (i.e., biaxial), or sensitive to vertical, horizontal and anteroposterior motion (i.e., triaxial).
Data collected by an actigraph is measured in one of three ways.
The “zero crossing” method counts the number of times a signal’s voltage rises above or falls to zero volts in an epoch; this method measures the frequency of a signal but not its intensity.
The “time above threshold” method measures the length of time within an epoch that a signal’s voltage is above a set threshold; this method does not measure the intensity of the signal.
The “proportional integrating measure” method measures the length of time a signal is above a set threshold, as well as its amplitude (i.e., strength). Once the movement data in an epoch are collected, the algorithm is used to determine whether the epoch indicates wake or sleep and whether the person is in bed or out of bed.
Assessing sleep disturbances in people with Parkinson’s disease can be problematic when using actigraphy because algorithms in current use may misinterpret dysfunctional motor activity such as tremors, bradykinesia, dyskinesia, and restricted arm movement during walking or drug-induced hypermotility. For this reason, scientists have worked to develop an actigraph that can more accurately assess sleep-wake cycles in people with Parkinson’s disease.
For example, in a 2013 study comparing actigraphy and PSG in patients with mild to moderate Parkinson’s disease, researcher Maglione and colleagues found that, when using the actigraph manufacturer’s threshold settings of 20, 40 or 80 (i.e., the number of activity counts within an epoch; values above the threshold indicate wake and values below the threshold indicate sleep), no setting was ideal for determining the time in bed, total sleep time, wake after sleep onset and sleep efficiency. The researchers further found that when they changed from the manufacturer’s threshold of 10 immobile minutes or 10 mobile minutes as an indicator of sleep onset or wake, respectively, to five immobile minutes and five mobile minutes, they could estimate sleep onset in these patients. Based on these results, Maglione proposed using a low-activity threshold to assess sleep parameters in patients with mild to moderate Parkinson’s disease.
In 2012, Griffiths and colleagues4 reported developing an algorithm that detected bradykinesia and dyskinesia in patients with Parkinson’s disease. The algorithm is used in the Parkinson’s KinetiGraph (Global Kinetics Corporation, Melbourne, Australia), which is a device worn on the wrist like an actigraph. The Parkinson’s KinetiGraph has shown promising results in studies.
In 2014, Kotschet and colleagues5 were the first scientists to report using the Parkinson’s KinetiGraph to detect bradykinesia and dyskinesia to assess daytime sleepiness in people with Parkinson’s disease. The Parkinson’s KinetiGraph is an actigraph-like device that detects bradykinesia as movement with low acceleration and amplitude and a long interval between movements, and detects dyskinesia as movement with a normal amplitude and normal acceleration but short period without movement.4 The device produces a bradykinesia score (BKS) every two minutes, which corresponds to four 30-second polysomnograph epochs. A score less than 80 indicated the person was most likely awake and a score greater than 80 indicated sleep.
In the Kotschet study, healthy individuals without Parkinson’s disease (i.e., controls) and individuals with Parkinson’s disease wore the device continually for 10 days. The proportion of time immobile during the daytime was higher in patients with Parkinson’s disease than in the controls (i.e., the patients were experiencing sleepiness during the daytime). The researchers further found that, in 53 percent of patients, the proportion of time immobile increased within 30-60 minutes after they took their daytime dose of levodopa, which indicated that treatment increased daytime sleepiness.
Sarah McGregor and colleagues7 used the Parkinson’s KinetiGraph to assess immobility and sleep stages during nocturnal sleep. They used a BKS less than 80 as an indicator that a person was awake or in N1 sleep and a BKS greater than 110 as an indicator that a person was in N2, N3 or REM sleep. Study participants without Parkinson’s disease underwent simultaneous PSG and Parkinson’s KinetiGraph recordings. A comparison of PSG epochs with the BKS (two-minute epoch) revealed that a BKS less than 40 indicated a person was awake, a BKS of 40-80 indicated a person was between wake and N1 sleep, a BKS of 80-111 indicated a person was between N1 and N2 sleep, and a BKS greater than 110 indicated a person was in N2 sleep.
Based on the polysomnography findings, the sleep studies were classified as “normal” (i.e., normal sleep study), “normal minus” (i.e., normal sleep study with increased leg movements, changes in oxygen saturations, or sleep fragmentation) or “abnormal” (i.e., abnormal sleep study). Bradykinesia scores indicating wake and sleep were used to estimate a person’s percent time awake, percent time asleep, percent time immobile, sleep quality and median fragment length. The BKSs revealed no differences between normal and abnormal sleep when assessing each of these parameters separately. However, compared with polysomnographic data, the sum of the BKSs of three parameters — percent time sleeping, percent time immobile and sleep quality — allowed the researchers to distinguish between normal and abnormal nocturnal sleep.
Accurate information regarding a patient’s sleep and wake cycles can be used to determine treatment response, the best time to administer a medication and whether changes in symptoms are occurring. Objective information provided by a device such as the Parkinson’s KinetiGraph would be particularly useful in patients who may not recognize a change in symptoms has occurred (e.g., increased daytime dozing) and therefore may not communicate this to a clinician. For now, scientists continue working to improve the technology.
- Sadeh A, Sharkey KM, Carskadon MA. Activity-based sleep-wake identification: An empirical test of methodological issues. Sleep. 1994;17:201-207.
- Cole RJ, Kripke DF, Gruen W, et al. Automatic sleep/wake identification from wrist activity. Sleep. 1992;15:461-469.
- Tahmasian M, Khazaie H, Sepehry AA, et al. Ambulatory monitoring of sleep disorders. Journal of Pakstan Medical Association. 2010;60(6):480-487.
- Griffiths RI, Kotschet K, Arfon S, et al. Automated assessment of bradykinesia and dyskinesia in Parkinson's disease. Journal of Parkinson's Disease. 2012;2:47-55.
- Kotschet K, Johnson W, McGregor S, et al. Daytime sleep in Parkinson's disease measured by episodes of immobility. Parkinsonism and Related Disorders. 2014;20:578-583.
- Klingelhoefer L, Rizos A, Sauerbier A, et al. Night-time sleep in Parkinson's disease—the potential use of Parkinson's KinetiGraph: A prospective comparative study. European Journal of Neurology. 2016;23:1275-1288.
- McGregor S, Churchward P, Soja K, et al. The use of accelerometry as a tool to measure disturbed nocturnal sleep in Parkinson's disease. NPJ Parkinson's Disease. 2018;4:1 doi: 10.1038/s41531-41017-40038-41539.