Most actigraphs are wristwatch-sized devices that use a piezoelectric accelerometer to measure activity levels of the person wearing the device. The use of motion as a measure is common in medical devices. For example, pacemakers have a mechanism that adjusts the rate in response to the patient’s activity level, whereas pedometers use motion as a way to monitor the number of steps a person takes. The use of activity monitoring to estimate sleep-wake patterns has been around for many years (4). Early devices were self-contained activity counters with integrated circuits and memory that provided off-line data retrieval (5). Over the years, actigraphs have become more compact and lightweight, with greater memory for data storage and a longer battery life (3). In addition to the newer models of actigraphs being noninvasive and compact, many include a watch capability on the device, so it is also functional to the user. These devices do not interfere with everyday activities, and most brands are also water resistant or even waterproof.

In 2018, the American Academy of Sleep Medicine (AASM) published updated guidelines for the use of actigraphy in the assessment of sleep and sleep disorders (6), and in 2014, the International Classification of Sleep Disorders, 3rd edition provided additional details about the use of actigraphy for the diagnosis and treatment of sleep disorders (7). Specifically, actigraphy should be used (a) for a minimum of 7 days before a multiple sleep latency test to ensure sufficient sleep duration and an appropriate sleep schedule; (b) for a minimum of 7 days with unrestricted sleep opportunity to verify idiopathic hypersomnia (with a 24-hour total sleep time ≥660 minutes); (c) to verify insufficient sleep syndrome if there is doubt about the accuracy of a clinical history or sleep diaries; and (d) to demonstrate an irregular circadian rhythm (i.e., delayed sleep-wake phase, advanced sleep-wake phase, and non-24-hour sleep-wake rhythm).

Actigraphy provides many benefits over PSG. During an overnight PSG, patients often have to travel to a center that may be at a distance from their home. In addition, they are sleeping in an unfamiliar setting, removed from the comfort of their own beds, and required to wear numerous sensors that are placed on the head and body. The sensations from the sensors may be overstimulating and difficult to tolerate for some patients (e.g., children with autism spectrum disorders or older adults). Furthermore,
PSG most often records only one night of sleep, which may not be representative of the person’s typical sleep because of the “first-night effect,” (8). Actigraphy, in contrast, can be worn in the patient’s natural home/sleep environment and for an extended period.

However, actigraphy is not a substitute for PSG. PSG uses multiple channels to measure the different physiologic stages of sleep as well as the respiratory state of the patient during the night. These measurements are essential in accurately diagnosing sleep-related breathing disorders, movement-related sleep disorders, and narcolepsy. Actigraphy may be able to measure the disruptions in sleep caused by sleep-related breathing disorders or movement-related sleep disorders; however, it cannot accurately identify the causes of these awakenings. Information from actigraphy can be a useful tool in clinical and research settings, but it should not be used as a simple substitute for PSG.

Activity counts are measured per epoch in actigraphy. Although some devices have a fixed epoch length of 1 minute, other devices allow the user to set the epoch length to 15 seconds, 30 seconds, 1 minute, or 2 minutes. Each device has a different way to digitize the signal. For some devices, the user can choose the data mode; for other devices, the data mode is automatically selected. When the user can select the data mode, there are three choices: time above threshold (TAT), zero crossing mode (ZCM), and proportional integration mode (PIM). TAT measures the amount of time per epoch the activity waveform is above the set threshold; ZCM measures the number of times per epoch the waveform crosses a threshold set close to zero; PIM, also known as digital integration, measures the area under the curve for each epoch (see reference [3, 5] for more information). Newer devices allow for multiple modes to be used simultaneously. Different studies have provided support for different measurement modes (1). Owing to the variance in results, it is questionable whether there is one best method of measuring the digitalized signal from the actigraphy. Further study should be conducted on a wider range of user populations to see if one mode or a combination of modes yields the most accurate results.

When the actigraphs are returned by the patient, the devices are downloaded to a computer using a device-specific interface. Each brand of actigraph has proprietary software that uses device-specific scoring algorithms (e.g., Sadeh and Cole-Kripke) or wake sensitivity (e.g., medium, high, and low) thresholds to determine whether each epoch should be scored as sleep or wake. The selection of which algorithm to use should be based on the sample being studied (3, 9, 10). Once the data have been scored, the program will generate an actigram (a picture of the patient’s sleep during the study; see Figs. 45-1 and 45-2 for examples) as well as summary variables such as total sleep time, actigraphic sleep-onset time, wake after sleep onset, and actigraphic sleep-offset time. Variables such as bedtime, wake time, and sleep-onset latency should be reported only if the patient also keeps a concurrent daily sleep diary (discussed further later) or uses the event marker feature included with some devices. Along these lines, sleep efficiency (SE) should be calculated only if a diary is included (providing bedtime and wake time; SE = minutes of sleep from actigraphic sleep onset to actigraphic sleep offset/minutes from bedtime to wake time × 100).

Although actigraphy has several benefits over PSG (e.g., relatively low cost, nonintrusive device that can be used for extended periods to estimate sleep-wake patterns), there are several concerns about validity that should be considered. The two primary areas of concern are (1) specificity to detect wake and (2) artifact.