ACTIGRAPHY Wactisleep-BT Monitor Micro Motionlogger Watch Motionlogger Watch Motionwatch 8 Actiwatch Spectrum Somnowatch

Company ActiGraph Ambulatory Monitoring Inc CamNtech Philips Respironics SOMNOmedics America Inc

Product Name ACTIGRAPHY wActiSleep-BT Monitor Micro Motionlogger Watch Motionlogger Watch MotionWatch 8 Actiwatch Spectrum SOMNOwatch www.ambulatory-monitoring. www.ambulatory-monitoring. www.somnomedics- Website www.actigraphcorp.com www.camntech.com www.philips.com/actigraphy com com diagnostics.com 1-year standard warranty; Cost/Warranty $299/1 year 1 year 1 year $750/2 years 2 years extended warranty available 3-axis solid state Miniaturized poly-silicone-surface Accelerometer accelerometer with digital Solid state triaxial Solid state triaxial MEMS – 3 axis Solid-state “piezoelectric” micromachined Technology filtering sensor (MEMS) 3.81 cm x 2.54 cm x 1.016 Product Dimensions 4.6 cm x 3.3 cm x 1.9 cm 3.6 cm x 3.6 cm x 1.2 cm 5.5 cm x 4.5 cm x 1.8 cm H 4.8 cm x 3.7 cm x 1.4 cm 4.5 cm x 4.5 cm x 1.6 cm cm Weight, Including 22 grams 30 grams 65 grams 16.8 grams 30 grams 30 grams Batteries and Wrist Band Rechargeable lithium Li-Ion-Accu, inbuilt, Batteries 1 DL2430 disposable coin cell 1 DL2450 disposable coin cell Common watch R2032 Lithium coin cell CR 2430 polymer rechargeable 1-year battery life (with continuous Battery Life 31 days 30+ days 30+ days 1 year 26+ days use at 1-minute epoch) Type of Memory Storage 2 GB nonvolatile 2 MB nonvolatile 2 MB nonvolatile 4 MB nonvolatile 2 MB nonvolatile 8 MB internal storage card Recording Time at 1- 36 days—Activity plus 120 days 30+ days 30+ days 182 days 45 days Minute Sample Interval 4 channels of light data Collects as raw data (30- Sampling rate 1 per second up to 100 Hz), processing in 1, 15-second, 30-second, 1, 2, 5, 10, 15, 30, 60 256 per second; storage rate 1 Logging Intervals 2, 3, 5,10, 15, 30, 60, 120, 1, 2, 5, 10, 30, 60 seconds 2, 5, 15, 30, 60 seconds 1-minute, 2-minute, seconds per 120 seconds (2 minutes) up to 150, 180, and 240-second 5-minute 256 per second intervals Red, green, and blue color Light Sensor Type Photodiode Photodiode Photodiode Digital NPN-phototransistor sensitive photodiodes blue 400-500 nm; 450 to 700 nm (600 nm Wavelength Range 400-700 nm (520 nm peak) 400-700 nm (520 nm peak) 0 to 64,000 lux green 500-600 nm; 350-970 nm peak) red 600-700 nm Immersible 1 m for 30 min- Water-resistant (waterproof on Moisture Protection Water-resistant 100 m water-resistant 50 m water-resistant Waterproof utes per IPX7 IEC 60529 demand) Operating Temperature -4°F to 140°F 33ºF to 120ºF 33ºF to 120ºF 0°F to 150°F 41°F to 104°F 41°F to 104°F Range Infrared to dock 57.6 kpbs; Communications USB, Bluetooth LE USB-to-IrDA USB-to-IrDA USB USB to PC (USB 1.1 and 2 USB Interface compatible) Actiware: Windows 2000, Windows XP, or Windows Vista; Windows XP, Windows Computer Operating Windows XP, Windows Vista, Windows XP, Windows Vista, Windows XP, Windows 7, Actiware CT: Windows 2000, Windows XP, 7, and other Vista, Windows 7, Windows Windows 7, Windows 8 Windows 7, Windows 8 Windows 8 Windows XP Professional, Windows operating systems System Compatibility 8, Parallels Desktop for Mac Windows Vista Business, or Windows Vista Ultimate Peer Reviewed Yes Yes Yes Yes Yes Yes Validation Articles Device also records in a red, green, and blue light only Integrated wear time sensor mode; off-wrist detection; Adjustable time of day/date for off wrist detection, sup- event marking; shock and display, backlight, event Event marker button, sleep/ port for heart rate (wireless vibration resistant; dust, heat, marker with visual/audible wake analysis, PLM scoring, 1 HR sensor required), PLM Time/date display, event and perspiration resistant; feedback, visual status and external channel possible (2nd scoring, automatic bedtime marker with LED feedback, The Event Marker button has latex/nickel free; biocompat- battery indicators, multimode Actisensor, ECG, EEG, respiratory), detection, body position, visual status indicator, a visual LED feedback indica- ibility tested; adheres to the Additional Features data collection, temperature programmable start and stop and daytime activity profile. multimode data collection, tor, confirming the event was requirements of 21 CFR Part channel, off-wrist detection periods for several recording Compatible mobile app for temperature channel, off- marked 820, and ISO 9001 and 13485 channel, PVT test, user rating periods, linking of several iPhone, iPad, and Android wrist detection. certification; programmable scale entry, alarms (1 user SOMNOwatch recordings thanks to platforms supports real time time/date display with back- and up to 10 fixable), stop- high synchronization rate device communication and light; status feedback; multiple watch, countdown timer data reports. light and activity data collection modes; NIST traceable calibra-

Information for this guide based on data submitted by product manufacturers. Sleep Review strives for accuracy in all data but cannot be held responsible claims made by manufacturers. tion; sleep mode

10 • sleepreviewmag.com DECEMBER 2013 Company ActiGraph Ambulatory Monitoring Inc CamNtech Philips Respironics SOMNOmedics America Inc

Product Name

wActiSleep-BT Monitor Micro Motionlogger Watch Motionlogger Watch MotionWatch 8 Actiwatch Spectrum SOMNOwatch www.ambulatory-monitoring. www.ambulatory-monitoring. www.somnomedics- Website www.actigraphcorp.com www.camntech.com www.philips.com/actigraphy com com diagnostics.com 1-year standard warranty; Cost/Warranty $299/1 year 1 year 1 year $750/2 years 2 years extended warranty available 3-axis solid state Miniaturized poly-silicone-surface Accelerometer accelerometer with digital Solid state triaxial Solid state triaxial MEMS – 3 axis Solid-state “piezoelectric” micromachined Technology filtering sensor (MEMS) 3.81 cm x 2.54 cm x 1.016 Product Dimensions 4.6 cm x 3.3 cm x 1.9 cm 3.6 cm x 3.6 cm x 1.2 cm 5.5 cm x 4.5 cm x 1.8 cm H 4.8 cm x 3.7 cm x 1.4 cm 4.5 cm x 4.5 cm x 1.6 cm cm Weight, Including 22 grams 30 grams 65 grams 16.8 grams 30 grams 30 grams Batteries and Wrist Band Rechargeable lithium Li-Ion-Accu, inbuilt, Batteries 1 DL2430 disposable coin cell 1 DL2450 disposable coin cell Common watch R2032 Lithium coin cell CR 2430 polymer rechargeable 1-year battery life (with continuous Battery Life 31 days 30+ days 30+ days 1 year 26+ days use at 1-minute epoch) Type of Memory Storage 2 GB nonvolatile 2 MB nonvolatile 2 MB nonvolatile 4 MB nonvolatile 2 MB nonvolatile 8 MB internal storage card Recording Time at 1- 36 days—Activity plus 120 days 30+ days 30+ days 182 days 45 days Minute Sample Interval 4 channels of light data Collects as raw data (30- Sampling rate 1 per second up to 100 Hz), processing in 1, 15-second, 30-second, 1, 2, 5, 10, 15, 30, 60 256 per second; storage rate 1 Logging Intervals 2, 3, 5,10, 15, 30, 60, 120, 1, 2, 5, 10, 30, 60 seconds 2, 5, 15, 30, 60 seconds 1-minute, 2-minute, seconds per 120 seconds (2 minutes) up to 150, 180, and 240-second 5-minute 256 per second intervals Red, green, and blue color Light Sensor Type Photodiode Photodiode Photodiode Digital NPN-phototransistor sensitive photodiodes blue 400-500 nm; 450 to 700 nm (600 nm Wavelength Range 400-700 nm (520 nm peak) 400-700 nm (520 nm peak) 0 to 64,000 lux green 500-600 nm; 350-970 nm peak) red 600-700 nm Immersible 1 m for 30 min- Water-resistant (waterproof on Moisture Protection Water-resistant 100 m water-resistant 50 m water-resistant Waterproof utes per IPX7 IEC 60529 demand) Operating Temperature -4°F to 140°F 33ºF to 120ºF 33ºF to 120ºF 0°F to 150°F 41°F to 104°F 41°F to 104°F Range Infrared to dock 57.6 kpbs; Communications USB, Bluetooth LE USB-to-IrDA USB-to-IrDA USB USB to PC (USB 1.1 and 2 USB Interface compatible) Actiware: Windows 2000, Windows XP, or Windows Vista; Windows XP, Windows Computer Operating Windows XP, Windows Vista, Windows XP, Windows Vista, Windows XP, Windows 7, Actiware CT: Windows 2000, Windows XP, 7, and other Vista, Windows 7, Windows Windows 7, Windows 8 Windows 7, Windows 8 Windows 8 Windows XP Professional, Windows operating systems System Compatibility 8, Parallels Desktop for Mac Windows Vista Business, or Windows Vista Ultimate Peer Reviewed Yes Yes Yes Yes Yes Yes Validation Articles Device also records in a red, green, and blue light only Integrated wear time sensor mode; off-wrist detection; Adjustable time of day/date for off wrist detection, sup- event marking; shock and display, backlight, event Event marker button, sleep/ port for heart rate (wireless vibration resistant; dust, heat, marker with visual/audible wake analysis, PLM scoring, 1 HR sensor required), PLM Time/date display, event and perspiration resistant; feedback, visual status and external channel possible (2nd scoring, automatic bedtime marker with LED feedback, The Event Marker button has latex/nickel free; biocompat- battery indicators, multimode Actisensor, ECG, EEG, respiratory), detection, body position, visual status indicator, a visual LED feedback indica- ibility tested; adheres to the Additional Features data collection, temperature programmable start and stop and daytime activity profile. multimode data collection, tor, confirming the event was requirements of 21 CFR Part channel, off-wrist detection periods for several recording Compatible mobile app for temperature channel, off- marked 820, and ISO 9001 and 13485 channel, PVT test, user rating periods, linking of several iPhone, iPad, and Android wrist detection. certification; programmable scale entry, alarms (1 user SOMNOwatch recordings thanks to platforms supports real time time/date display with back- and up to 10 fixable), stop- high synchronization rate device communication and light; status feedback; multiple watch, countdown timer data reports. light and activity data collection modes; NIST traceable calibration; sleep mode Behavioral Sleep Medicine, 13:S4–S38, 2015 Copyright © Taylor & Francis Group, LLC ISSN: 1540-2002 print/1540-2010 online DOI: 10.1080/15402002.2015.1046356

The SBSM Guide to Actigraphy Monitoring: Clinical and Research Applications

Sonia Ancoli-Israel, PhD Departments of Psychiatry and Medicine, University of California, San Diego Jennifer L. Martin, PhD David Geffen School of Medicine at the University of California, Los Angeles VA Greater Los Angeles Healthcare System, Geriatric Research, Education and Clinical Center Terri Blackwell, MA Research Institute, California Paciﬁc Medical Center, San Francisco

Luis Buenaver, PhD Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine Lianqi Liu, MD Department of Psychiatry, University of California, San Diego Lisa J. Meltzer, PhD Department of Pediatrics, National Jewish Health, Denver Avi Sadeh, DSc School of Psychological Sciences, Tel Aviv University, Israel Adam P. Spira, PhD Department of Mental Health, Johns Hopkins Bloomberg School of Public Health Department of Psychiatry and Behavioral Sciences, Johns Hopkins School of Medicine Daniel J. Taylor, PhD Department of Psychology, University of North Texas

EDITORS: Judith A. Owens, MD, MPH Michael T. Smith, PhD, CBSM Harvard Medical School Johns Hopkins School of Medicine

S4 SBSM GUIDE TO ACTIGRAPHY MONITORING S5

CHAPTER 1 Introduction

Actigraphs are compact, lightweight, computerized accelerometer-based devices that have been used to evaluate sleep and wake in humans for nearly four decades.1 While polysomnography (PSG) continues to be the gold standard for recording sleep, wrist actigraphy offers the advantages of being easier to use and less expensive and cumbersome. Although actigraphy should not be viewed as a substitute for PSG when an overnight sleep study is indicated (e.g., when EEG parameters are needed or sleep disorders requiring PSG are suspected), actigraphy allows for extended continuous recording of both nocturnal and daytime sleep periods for days or weeks in the patient’s home sleep environment. Actigraphy can therefore yield information that is not captured during a single night in the sleep laboratory or via portable ambulatory monitoring. These features are particularly advantageous in select patient groups such as those with suspected circadian rhythm sleep disorders or insomnia complaints, in pediatric patients with sleep difficulties, and in older adults. Actigraphy can also be useful for objectively assessing the impact of clinical interventions such as cognitive behavioral treatment of insomnia (CBT-I) or changes in schedule, such as adjustment to shift work. In addition, actigraphy can provide information about an individual’s sleep patterns and sleep duration prior to quantifying daytime sleepiness with the multiple sleep latency test (MSLT) in order to ensure that the patient’s sleep was typical (i.e., not unusually restricted or extended) during the week prior to the MSLT. While not indicated for the diagnosis of medically based sleep disorders such as obstructive sleep apnea (OSA) and restless legs syndrome (RLS), actigraphy can also be used to reliably assess other sleep parameters in these patients; for example, one study found that when subjective data were combined with actigraphy data in patients with sleep-disordered breathing, total sleep time and sleep efficiency did not differ from PSG.2 When actigraphy first became available, it was used primarily for research purposes, as many of the methodological considerations had not yet been adapted for clinical applications. Since then, there have been numerous studies investigating the use of actigraphy both in research settings and in clinical populations. The review articles written over the last several years have consistently documented that wrist actigraphy can reliably estimate sleep and wake across the lifespan, and is an appropriate and useful addition to the clinical evaluation and assessment of treatment benefits in patients with specific sleep disorders, particularly insomnia, circadian sleep/wake disturbances, and periodic limb movement disorder (PLMD).1;3–5 Actigraphs have evolved considerably since they were first introduced in the early 1970s6–8; currently available devices have both sophisticated accelerometers to measure movement and sufficient memory to record and store data for multiple weeks. Software programs have also improved over time and now allow for reliable automatic scoring based on established algorithms. They also yield data reports and summaries that provide useful information for clinical and research applications. Many software packages also have scoring programs for circadian rhythm activity analysis, further enhancing the utility of actigraphs for the evaluation of select patients. The Centers for Medicare Services (CMS) designated a Current Procedural Technology (CPT®) code for actigraphy monitoring (code 95803) defined as “actigraphy testing, recording, S6 ANCOLI-ISRAEL ET AL. analysis, interpretation, and report (minimum of 72 hours to 14 consecutive days of recording),” enabling providers to apply for reimbursement in clinical settings.9 As the use of this technology continued to expand, it became increasingly clear that the field was in need of a manual to standardize the use of actigraphy devices, and to define methods for data collection and scoring and interpretation of results. A more systematic manualized approach would further enhance the use of actigraphy in clinical and research settings and allow for comparability across providers, investigators, and settings. Responding to this unmet need, this scoring and instructional manual was commissioned by the Society of Behavioral Sleep Medicine (SBSM) primarily to assist clinicians, but in addition to inform researchers in the use of actigraphy. The content was developed by experts in the field of actigraphy and sleep, and represents the current “state of the science” in the use of wrist actigraphy for clinical populations. The manual chapters cover a wide variety of core topics, including minimal technical specifications for devices and software, patient instructions, auxiliary clinical information that must be collected to optimize the accuracy of recordings, editing and scoring of actigraphy data, interpretation of results, and report generation. A brief discussion about the future of actigraphy as a clinical and research tool is included. When appropriate, some chapters include a section at the end highlighting any special considerations for research applications that may differ from routine clinical implementation of actigraphy.

CHAPTER 2 Minimal Technical Specifications to Consider When Purchasing an Actigraph

This chapter helps clinicians and researchers understand the key features to evaluate when choosing an actigraph (summarized in Table 1). It should be noted that the devices discussed here are those specifically designed and validated for measuring sleep-wake patterns; devices currently on the market that were developed for direct-to-consumer sale typically do not incorporate an accelerometer or a sleep-scoring algorithm that has been validated for sleep detection, and thus are not included. While these relatively low-cost devices have a number of appealing features, including integration with mobile devices and software, insufficient data exist at the present time to establish the validity and reliability of sleep parameters measured by these devices. In fact, the few available studies suggest they are not sufficiently accurate in either clinical or research settings, and thus their use cannot currently be recommended.10;11 The number and types of features available on actigraphy devices vary; and their relative importance and desirability for the intended purpose should be carefully considered when comparing specific models. These may include the sampling rate of the accelerometer, the modes available for calculating activity data, appearance, size and weight, additional functions such as the presence of an event marker button or the recording of ambient light data, battery life, data storage and computer system requirements, safety concerns, and water resistance. It should be emphasized that the authors and editors of this manual do not endorse any specific SBSM GUIDE TO ACTIGRAPHY MONITORING S7

TABLE 1 Technical Specifications for Actigraphy Devices

Type Deﬁnition

Accelerometers Omnidirectional or triaxial accelerometers with available validation studies are required. The ability to collect data as “time above threshold” or “digital integration” mode is recommended. The ability to record data in 30-second or 1-minute epochs is required. Flexible data collection parameters (e.g., epoch length) are recommended. Appearance Productdimensionsshouldbe considered, particularly among infants and young children, as the weight of some devices may be cumbersome or limit adherence. Event markers Eventmarkersare a useful additionto an actigraphy log. These are recommended. Light sensors Light sensorsembeddedin the device can facilitate identiﬁcation of bed times and rise times. These are recommended. Battery type A battery (rechargeableordisposable)that can record data for a minimum of 3 days is required. A battery (rechargeable or disposable) that can record data for a minimum of two weeks is recommended. Data storage and Nonvolatile memory that allows information to be stored even if the battery fails is memory recommended. The ability to store data for a minimum of 3 days is required. Customer support Support for both device and software is required.

brand or manufacturer, but rather provide the detailed information below in order to guide the user in selecting the most appropriate device.

MINIMUM REQUIREMENTS

Accelerometers Different brands of actigraphs use a variety of accelerometer technologies and some accelerometers may be more reliable or accurate than others, especially when considering the population that is to be studied (e.g., children vs. older adults). It is important to conﬁrm that the accelerometer data have been validated for each device being considered, with results published in peer-reviewed journals. Typically, omnidirectional or triaxial accelerometers are used for the measurement of sleep. These devices are different from accelerometers that are used to estimate calorie expenditure or to calculate pedometry variables. There are also multiple modes of collecting and calculating activity data, with the most common being time above threshold, zero crossing, and integration modes. Different manufac- tures may use different modes, with different names, and only some will give the user a choice in setting the mode for data acquisition. As described below, each of these modes determines sleep and wake based on a combination of three primary variables:

1. duration and/or frequency of movements 2. amplitude or strength of the movement 3. acceleration or speed of the movement S8 ANCOLI-ISRAEL ET AL.

Each mode has advantages and disadvantages. The time above threshold method counts the cumulative amount of time per epoch (a given time period) that the level of the movement is above some threshold (commonly 0.1 to 0.2 g). In some software packages the threshold can be adjusted (e.g., high, medium, or low). The disadvantage of this mode is that the amount the amplitude above threshold is ignored, as is the acceleration of the movements over time. The zero-crossing method counts the number of times per epoch that the activity signal crosses zero. The disadvantages of this method are that, as with the time above threshold, the amplitude of the movement is ignored, as is the acceleration of movements over time. In addition, high-frequency artifacts may be counted as movement. Digital integration involves calculating the area under the curve for the accelerometry output, which is sampled at a high rate for each epoch to be scored. The advantages are that this procedure involves rectifying the analog signal, doubling the amount of data available for analysis of each epoch; and the output reﬂects both acceleration and amplitude of movement. In this mode, however, the duration and frequency of movements are not evaluated. In studies that have compared the three modes, digital integration was found to be better than time above threshold for identifying movement amplitude and both digital integration and time above threshold functioned better than zero crossing mode for accurately detecting sleep and wake.12 It should also be noted that some actigraphs can record simultaneously in more than one mode, making it possible to determine which method to use after data are collected. Finally, research has shown that some modes, as well as scoring algorithms, are more accurate in younger populations while others are more accurate in older adults.13 It is therefore necessary to review published data on the speciﬁc patient population to be studied to help determine which method of data gathering is optimal.

Appearance The product dimensions (size, weight) and appearance vary across manufacturers. Some actigraphs have a digital display, while others do not. The digital display often includes a watch face (which may allow patients to wear the actigraph in place of a wristwatch), and some allow patients to view feedback from the actigraph recording itself (which may be useful in monitoring other behavioral markers along with sleep). The device selected should meet the needs of the speciﬁc patient group.

Additional Functions and Features Some models allow for the collection of adjunctive data. Since each of these parameters and features will impact battery life and memory consumption, the user manual that accompanies the device should be reviewed to identify the maximum recording duration given a particular set of parameters. Many models have an event marker button that can be useful in editing the data (e.g., for bedtime, time out of bed in the morning, times the device was removed). Subjective data such as ratings of fatigue or sleepiness can be collected with some models. Light sensors are often available, and light data (e.g., sunrise, sunset, lights out at night) can be useful in editing the actigraph recordings as well as for determining overall light exposure. If light data recording is desired, consideration should be given to what levels of light exposure SBSM GUIDE TO ACTIGRAPHY MONITORING S9 measurement are needed, as different devices can record different wavelengths and some specify minimum and maximum light levels that can be detected. Some devices include an actual watch face, and thus settings are available for date and time, including time zone and 24-hour clock display. Some units allow patients to have control over changing the visible time/date, or disabling the clock feature altogether. Finally most actigraphs are water-resistant and do not need to be removed for bathing (see chapter 3 for more information regarding water exposure and immersion).

Battery Options Varieties of battery options are available. Some batteries are rechargeable, while others require routine replacement. Battery life should be considered when selecting a device because the unit should be able to record for the entire length of time during which a patient would generally be required to wear it. Some devices have a long battery life (up to several months), but need to be shipped back to the manufacturer for battery replacement. The appropriateness of each power source should be considered, given the planned use of the device.

DATA STORAGE AND MEMORY

Epoch Length Data are collected and stored in epochs, which can range from 1 second to 5 minutes and vary by device. The epoch length is sometimes modifiable by the user. The longer the epoch length, the less the memory and battery life are used. However, longer epoch lengths decrease sensitivity and specificity to detect sleep and wake after sleep onset. The most validated and commonly used epoch lengths are 30 seconds and 1 minute. The size and type of memory storage can also vary. The amount of memory should allow for a length of recording time that is sufficient for the patient’s needs, at the desired epoch length. Nonvolatile memory allows for data retrieval even if the battery dies during the recording period.

Actigraph Interface With the Computer Currently, actigraphs need to be connected to a computer to initialize and to download data, and wireless technologies are not widely available. Communication interfaces vary and require different modes of connection to the computer. Some require a docking station device or cable (e.g., USB, infrared) for connection to the computer, sometimes requiring an additional power source or battery for the docking station as well. Several devices now have connections that are Bluetooth® compatible, an attractive alternative likely to become more common over time. Consider where and how a given device needs to be initialized and data downloaded when purchasing an actigraphy system. More information on interfaces can be found in chapter 3. S10 ANCOLI-ISRAEL ET AL.

SAFETY

While actigraphs are generally considered low-risk medical devices, there are some safety considerations that should be kept in mind. For example, the exterior of the actigraph that will be in direct contact with skin may contain substances which can cause irritation or to which some patients are allergic (i.e., latex or nickel). However, since an actigraph measures movement, direct contact with skin is not necessarily required; therefore the device could be worn over clothing or a cloth wristband to reduce the likelihood of skin rashes and allergic reactions. Some devices, however, use an off-wrist detection technology that does require skin exposure. There may also be specific restrictions on use that impact particular patient populations; for example, it should be determined whether the actigraph may be worn in the presence of radiation, magnetic fields, defibrillators, pacemakers, or oxygen actively in use.

SOFTWARE AND SCORING ALGORITHMS

The computer operating system used must be compatible with the actigraph software, and this may limit the use of some software systems in institutional settings. The amount of computer memory and storage space needed for the software and storage of actigraphic recording ﬁles should be appropriate for the computer system to be used; however, current computer storage parameters are typically ample for storing the relatively small actigraphy data ﬁles (typically

<1 megabyte per recording for week-long recording). The sleep-scoring algorithm employed by the device should have been validated and published in peer-reviewed articles, conﬁrming the accuracy of the sleep/wake assessment method used by the software compared to the gold standard of PSG. The availability of additional features for scoring or editing the ﬁles should be considered, as needed. These include the ability to:

View a visual actigram for hand scoring Switch from single-day display to dual-day or multiple-day display Change the start and end time of the display (e.g., midnight to midnight or 9 a.m. to 9 p.m.) Merge separate records Ability to readjust epoch length, such as from 15 seconds to 1 minute Force scoring of wake when it is clear that the device was off-wrist and the patient was not sleeping (for example, removed to play football); Replace a known wake time period with a certain level of activity (e.g., daily mean, or a ﬁxed number) Export epoch-by-epoch raw data Automatically score rest intervals and off-wrist detection algorithms

The sleep measures in the standard report output should be reviewed to insure they are acceptable (see chapter 6). Some software also allows for calculation of circadian activity rhythm parameters, so the user should determine if that feature is needed. SBSM GUIDE TO ACTIGRAPHY MONITORING S11

ADDITIONAL CONSIDERATIONS

There are many companies that provide actigraphs for clinical and research applications, and each type of device has both advantages and drawbacks. In addition to the points to consider listed above, as with any diagnostic device, the type of technical support package should be considered. The actigraph manufacturer should have a website or help line that can be contacted for troubleshooting problems with either the device or the software. Each device should have a unique serial number for tracking purposes, that is, linking the specific unit to the specific patient. This becomes especially important if it is anticipated that the patient will be monitored during several successive intervals in order to assess changes in sleep patterns over time, as the same actigraph device should be used throughout the entire monitoring period. Each device should offer a warranty; the cost and duration of these warranties vary by manufacturer. Some companies may offer the option to return the actigraph for refurbishment or recalibration on an annual basis, which may be desirable if the actigraph will be used for a long time period. As previously stated, a critical consideration is that the device chosen and its related sleep-wake algorithms must have been validated for the specific intended clinical or research purposes (age group, related medical issues, etc.).

CONSIDERATIONS FOR RESEARCH

While there may be situations in which clinical and research needs differ (for example, research may require additional parameters that are not necessarily needed for clinical applications such as recording periods lasting >1 month), the minimum requirements discussed above are essentially the same for both applications.

CHAPTER 3 Basic Patient Instructions for Use and Auxiliary Information

INITIALIZING/PROGRAMMING ACTIGRAPHS

Prior to the start of an actigraphy recording, devices must be set up, initialized, or programmed to initiate data collection. Although every device differs in the actual process, the ﬁrst step for all devices is to connect the actigraph to the computer through some type of interface (i.e., docking station: USB cable, infrared cable, Bluetooth® connection). Next, the parameters for the collection of data should be set using the software provided by the device manufacturer. Although the names and settings may differ by device, there are typical parameters that must be speciﬁed each time a device is set up for data collection. S12 ANCOLI-ISRAEL ET AL.

Data Collection Start and End Times Options typically include having data collection begin immediately or at some future time (e.g., 1 to 3 days later). The latter option is helpful in conserving battery life or memory capacity if recording will not begin immediately. Some devices also allow for a data collection endpoint to be set. Again, this can be used to conserve battery life in the event that the patient will maintain possession of the device long after the conclusion of the recording (e.g., patient cannot return for follow-up visit until several weeks after a 2-week recording).

Ambient Light Many devices have light sensors embedded in the actigraph itself, allowing for the collection of ambient light data. As mentioned previously, this can provide useful information for the scoring of actigraphy studies as change in light levels can be used to corroborate sleep diary- documented bedtimes and rise times. However, keep in mind, as mentioned above, that the use of the light sensor will decrease available memory and battery life, which may result in fewer recording days.

Rating Scale Capabilities

Some devices allow patients to provide a numeric rating for a predetermined clinical or research question to be answered in real time (e.g., “How much pain are you having right now?” or, “How sleepy do you feel?”).

PATIENT INSTRUCTIONS FOR WEARING AND USING ACTIGRAPH DEVICES

Patient Instructions It is essential to provide clear written and verbal instructions for the wearing and use of actigraph devices (see sample patient instructions in Appendix A). Individuals should be informed that the actigraph records movement, which is measured by an accelerometer similar to the accelerometer incorporated into many smart phones and personal fitness tracking devices. (Note: As delineated in chapter 2, “Technical Specifications,” we only recommend the use of actigraphs that have been established as valid and reliable tools for assessing sleep parameters. Much less is known about the validity of smart phone applications, or apps, that purport to record sleep/wake patterns. At the time of this writing, there is no evidence to support their validity for measuring sleep.) Showing patients an example of an actigram (Figure 1) will help them to understand the data that will (and will not) be collected. This example is also useful to clarify the importance of when to use an event marker and why it is critical to fill out the actigraphy log (both described below) in order to improve the accuracy of the results. SBSM GUIDE TO ACTIGRAPHY MONITORING S13

FIGURE 1 This graph represents the patient’s sleep during the study. The dates are on the left and the time is across the bottom, starting at noon, with midnight in the middle. The black sections are when the watch thought the patient was awake. The underlined sections are when the watch thought the patient was asleep. Shaded is the sleep period per actigraphy log. The event marker is the triangle at the top of the row. (Color ﬁgure available online.)

Device Placement and Bands Placement of the actigraph is an important consideration. Most devices are designed to be worn on the wrist. While the majority of studies have reported the validity of actigraph devices worn on the nondominant wrist to measure sleep-wake patterns, it has not been clearly determined whether wearing the device on the nondominant wrist yields more valid data than wearing it on the dominant wrist. While a few studies have recently examined the use of trunk or arm placement for wake and sleep in adults, this evidence remains limited. Thus, in adults, the nondominant wrist remains the best-validated placement of the device. In some patient populations with limited mobility (e.g., nursing home residents), however, the device may be placed on the dominant wrist so more movement can be captured. In contrast, in infants and toddlers the common practice is to attach the actigraph to the ankle rather than the wrist, in order to limit the child’s engagement with the device and promote safety. Most algorithms for these younger age groups have been based on ankle placement. There are also actigraph devices that have been designed speciﬁcally to be worn on the ankle to measure periodic limb movements in sleep in adults. S14 ANCOLI-ISRAEL ET AL.

The actigraph should ﬁt snugly around the wrist. It should not be so tight as to be uncom- fortable, but it should not be so loose that it moves freely around the wrist. For individuals who have concerns about wearing the device against their skin, a lightweight wristband can be placed under the device (after conﬁrming that an off-wrist detection feature will not erroneously indicate that the device is not being worn), or the actigraph can be placed on top of the sleeve. While this may impact readings for an off-wrist sensor, it will increase adherence to wearing the device while minimizing any issues related to skin sensitivities. Most actigraph devices come with a watchband that allows individuals to remove the device as desired. When a device is returned, these bands (as well as the device) should be thoroughly cleaned and disinfected prior to giving the device to another individual. Single-use disposable bands resembling hospital bracelets can also be used. Since these bands must be cut off, it will be apparent if the device has been removed during the recording; thus these types of bands are particularly useful if there are concerns about the individual removing the device and either forgetting to put it back on or losing it. The disadvantage of these types of bands, however, is that they limit the individual’s ability to remove the device for bathing, showering, and swimming, or if the individual will be engaging in an activity or is likely to be in an environment in which the device is likely to get damaged. If a device does not have a watch feature, and an individual prefers to wear his or her own watch, the participant’s wristwatch can either be worn next to the actigraph on the same wrist, or the participant may transfer the wristwatch to the dominant wrist. While ideally the actigraph is applied to the participant’s wrist by the clinician or researcher, this is not always possible (e.g., if the device is sent to participants by mail). In that case, detailed written instructions should be provided (supplemented by telephone instructions if possible). A graphic or photograph demonstrating proper placement can also be helpful.

Infants and children less than three years old and special populations. When working with children, there are two additional considerations for actigraph placement. First, as mentioned above, for infants and young children (under the age of 3 years), placement of the actigraph on the ankle has been well-validated and is preferred over wrist placement. Second, children with autism, sensory issues, or other neurodevelopmental disorders often will not tolerate the wrist placement of an actigraph device, and it can be placed on the ankle in these children as well. A pilot study has shown that placing the actigraph in a special pocket on the sleeve of a shirt is also a valid way to estimate sleep-wake patterns in pediatric patients.14 Although not as accurate as nondominant wrist placement, this alternative placement allows for data collection in individuals that would likely not otherwise tolerate this procedure.

Recording Length and Removing Devices During Study The number of days and nights that the actigraph will be worn will vary. According to CPT® code 95803, a minimum of 72 hours of recording is necessary to bill for actigraphy testing as a stand-alone service.9 Clinically however, 7 to 14 days of recording are more likely to provide adequate information about a patient’s sleep/wake patterns, as capturing both weekday and weekend sleep can help inform the clinical picture. To assess differences between weekday and weekend sleep patterns (essential for adolescents and others with variable sleep schedules), a 14-day recording that captures two weekends is preferred.15 SBSM GUIDE TO ACTIGRAPHY MONITORING S15

Individuals are typically instructed to wear the actigraph 24 hours per day for the entire assessment period, as continuous activity monitoring can help to place nocturnal movement data in context, and assist with interpretation of the findings. Even if nocturnal sleep is the main variable of interest, information about daytime napping behavior is often very helpful in rounding out the clinical picture and pinpointing problematic sleep patterns. A common situation, for example, involves a patient who dozes frequently during the day while presenting with sleep-onset insomnia. Without the daytime napping information, this type of clinical presentation could be misinterpreted. Furthermore, if actigraphy is being conducted to estimate circadian activity rhythms rather than just sleep, it is all the more important to keep the device on for as much of each 24-hour period as possible. However, this raises the important question of whether the actigraph should be removed when bathing or swimming, or engaging in other activities that would expose the actigraph to water or could potentially damage the device. Individuals should be informed whether the device they will be wearing is water-resistant, waterproof, or neither (see discussion in chapter 2). Briefly, waterproof devices may be kept on at all times including when swimming. Water-resistant devices need to be removed if submerged under water or exposed to moderate amounts of water. Devices that are neither water-resistant nor waterproof are to be removed in any situation in which water is involved. It must be emphasized to the patient that the actigraph should be replaced on the wrist as soon as the water-related activity is over. In addition, the time the device was removed and replaced on the wrist should be noted in the actigraphy log, along with any comments regarding what the patient was doing during this time (for example, swimming, which would suggest that the patient was awake during that time period and thus this interval should be manually scored as wake if the algorithm inappropriately scored this interval as sleep; see chapter 3). Another important point that should be conveyed to patients is that if an actigraph is removed, it should be reattached to the same wrist. In other words, an actigraph that was placed on the nondominant wrist should not be transferred or reattached on the dominant wrist and vice versa. Switching wrists during the assessment period may adversely impact data collection and interpretation, as activity levels may differ by wrist. For children and adolescents, concerns related to actigraph devices being lost or damaged at school, on the playground, and so forth need to be considered. A discussion with families about this possibility is essential in order to determine, for example, whether the child has physical education class or participates in sports during which the device could either be removed and placed in a locker, or might be damaged if not removed (e.g., football). In these cases, it is useful to proactively identify with the family specific behaviors that would be linked with the removal of the actigraph device before school, and the replacement of the actigraph device after school as reminders (e.g., remove device at breakfast and leave it on the kitchen table, and place the device back on the child’s wrist when the child is having an after-school snack).

Use of Light Sensors If the device being used contains a light sensor or photometer that will be activated during programming, it is important to inform individuals of this feature. They should be told that light exposure is an important synchronizer of sleep/wake patterns, that the actigraph will be recording the levels of light in their environment, and that they should be mindful that their S16 ANCOLI-ISRAEL ET AL. clothing does not cover the actigraph and thus obstruct the light sensor. Sleeves can be rolled up to prevent interference with light measurement, or in some cases, the device can be worn on top of lightweight clothing.

SUPPLEMENTAL DATA REQUIRED FOR SCORING AND INTERPRETATION

An advantage of actigraphs over self-report sleep measures is that these devices provide objective data about an individual’s sleep-wake patterns. Actigraphs provide information using automated scoring by software packages on sleep intervals throughout the 24-hour day. Pre- liminary estimates of sleep onset time (i.e., when the individual fell asleep), sleep offset time (i.e., terminal waking in the morning), and sleep duration (i.e., time from sleep onset to sleep offset) can be analyzed without additional information such as sleep diaries. However, for individuals with atypical sleep habits or patterns, sleep onset and offset times may not be accurately detected by the scoring software. In addition, supplemental information is always required to obtain sleep onset latency and sleep efficiency data (see below), as well as to identify misclassified sleep or wake in the recording (e.g., extended motionless time such as during a movie scored as sleep). This supplemental information can be provided through sleep diaries or logs, use of concurrent ambient light measurements, event markers on the actigraph device, or the use of daily phone calls. Whenever possible, more than one source of information (e.g., actigraphy log plus event marker) is preferable, allowing for better-informed scoring and interpretation of actigraphy results. In order to be able to calculate sleep onset latency and sleep efficiency, it is important to precisely identify the sleep period (e.g., lights off to lights on, or time person attempted to fall asleep to the time of getting up to start the day). Because “bedtime” can be interpreted in multiple ways, it is useful to have individuals identify both the time they got into bed and the time they attempted to fall asleep. For example, some individuals will read or watch television in bed prior to falling asleep. This time would not be counted within the sleep period (see chapter 5 on scoring and chapter 6 on interpretation for more information). Similarly, because some people spend time reading or watching television in bed in the morning, with no intention of sleeping, it can be helpful to assess the time that an individual wakes up and no longer is intending to sleep, as well as the time of getting out of bed. This is also useful for older patients who spend large amounts of time in bed or are bedridden for various reasons.

Actigraphy Log Unlike a detailed sleep diary, the actigraphy log often includes fewer variables, depending on the clinical or study question. For example, the consensus sleep diary16 is of great value when conducting CBT-I, but generally includes more information than what is needed for clinical or research applications of actigraphy. For that reason, even if the individual is ﬁlling out a sleep diary, an actigraphy log should also be completed if the following variables are not included in the sleep diary. The essential components for an actigraphy log are the following:

Time the individual got into bed Time the individual attempted to fall asleep at night SBSM GUIDE TO ACTIGRAPHY MONITORING S17

Time the individual woke up for the last time in the morning Time the individual got out of bed to start the day in the morning (no longer trying to sleep) Times of any daytime naps Times that the actigraph was removed from or replaced on the wrist Any unusual circumstances that might have affected sleep/wake patterns (e.g., illness, travel across time zones, novel sleeping environment, sitting still for prolonged periods of time such as during a movie)

Individuals should be encouraged to keep the actigraphy log in one speciﬁc and obvious place that will help them remember to complete the questions about daytime use each night and the questions about the night each morning (e.g., next to the bed). For children, it is important to consider that the actigraphy log is often completed by parents. However, adolescents should be encouraged to complete their own diaries, as parents are often not involved in their sleep routines and may not be aware, for example, of the time an adolescent goes to bed at night. Similarly, a child who is 8 years or older should be encouraged to help the parent complete the actigraphy log each day, as the child may be old enough to contribute valuable information in this domain. This is particularly important in regard to reporting prolonged sleep onset latency or nighttime awakenings of which the parent may not be aware.

Event Marker Some actigraphy devices include an event marker button, which the wearer is instructed to press to note a speciﬁc event. While the event marker should not be viewed as a substitute for an actigraphy log, it can be used to complement the log or to gather data about events such as device removals. Again, depending on the clinical question, individuals may be encouraged to press the event marker at certain times, including:

Time attempted to fall asleep at night During any nighttime awakenings Time awakened in the morning to start the day At the start or end of naps When the actigraph is removed and replaced

It is important to educate individuals about how to press the event marker (some models have a screen that shows whether the marker was properly pressed) and to explain that the event markers do not start or stop data collection; instead, pressing the event marker provides a time stamp on the recording that assists with the scoring and interpretation of the collected data (Note: Showing the patient a picture of an example actigram with event markers is helpful at this point; see Figure 1). Individuals should also be instructed not to press the event marker at times other than those speciﬁed. For example, some individuals may forget to press the event marker when they wake in the morning, and will then press it 3 hours later, believing that it is important for data collection. This can actually make interpretation of the event markers more challenging. S18 ANCOLI-ISRAEL ET AL.

Daily Phone Calls Although phone contact (or text messaging) may be cumbersome in clinical practice and thus is more often employed for research purposes, asking individuals to leave a voicemail (or text) at a designated number regarding the time they go to bed and wake up in the morning can provide valuable diary information. Not only does this provide regular information regarding the individual’s adherence to the study protocol, but it also provides a time-stamped record of bedtime and wake time that can further inform scoring and interpretation.

Electronic Sleep Diaries and Sleep Diary Apps The number of available sleep diaries that can be completed electronically is rapidly growing. While these are appealing for patients, they are typically not designed for use with actigraphy and thus they often do not capture all of the information needed for scoring and interpretation of actigraphy recordings (see above). As a result, these are not currently recommended unless they are speciﬁcally developed to measure the information required by actigraphy and the information can readily be made available to the technician scoring the actigraphy recording (i.e., the information can be gathered or transferred in ways compliant with the Health Insurance Portability and Accountability Act, or HIPAA).

CONSIDERATIONS FOR RESEARCH

Generally, the same guidelines regarding patient instructions and supplementary data apply for clinical and research applications. In research, however, the daily actigraphy log is often combined with a sleep diary completed by participants. Researchers may also wish to document speciﬁc activities of interest within the actigraphy log (e.g., meal times) and additional variables can be added as indicated by the research hypotheses to be tested.

CHAPTER 4 Identification of Invalid or Missing Data Periods and Preliminary Editing

To maximize the accuracy and validity of actigraphy recordings, speciﬁc steps are required for data cleaning by the individual scoring the record prior to interpretation. Immediately after the device is retrieved from the patient, the recording should be uploaded and reviewed. This process is required whether or not the software includes automatic scoring and rest-period detection. The actigraph log, as well as any other adjunctive data (e.g., event markers) should be used to assist the editing process. The main purpose of the initial review is to ascertain whether there are technical problems with the device or the recorded data. SBSM GUIDE TO ACTIGRAPHY MONITORING S19

Once the actigraphy log has been reviewed, the actigraphy ﬁle should be opened and data visually examined to get an overview of the record. Generally, there are two common types of issues that arise: device removals, and in pediatrics, artifacts associated with parenting (see Table 2). The following steps are recommended:

1. Omit times at the beginning and end of the recording when the device was not on the patient’s wrist: a. Edit the start of the file: Any time period in the beginning of an actigraphic recording during which the device was not worn should be deleted (note that data are typically hidden, rather than permanently deleted in most software packages). This will happen if the actigraph is initialized to begin recording before the patient actually puts it onto his or her wrist. If there are data from unwanted day(s) or night(s) before the official start of data collection, these too should be deleted. Most systems preserve and hide deleted data so that in the event that an error is made during this process, the data can be restored. Prior to initiatingthis step, it is important to note the specific requirements of the software being used. b. Edit the end of the file: Any blank time period at the end of an actigraphic recording should similarly be deleted. This will happen at the end of the recording period if the patient continues to wear the actigraph, if the actigraph is recording while in transit to clinic, or if the actigraph is removed some time before the data are downloaded. If there are any data from unwanted days or nights at the end of the data collection period, these should also be deleted. 2. Identify missing data: The most common issue encountered in the review of actigraphy data are periods with no activity in which the scorer cannot determine whether the patient was asleep or awake. If the actigraph was removed, the recording would show a flat line with no activity (i.e., activity values are at 0). During these time periods, the data typically need to be “deleted” from the analysis manually so they are not included as time asleep.

TABLE 2 Considerations for the Initial Data Review

Type Deﬁnition

Device removal When the actigraph is removed,the actigraphy log will help determine if the patient was awake during this time period (e.g., if patient reports swimming then it can be assumed he or she was awake and the scorer can manually score recording to reﬂect wakefulness rather than sleep). Parenting artifact For pediatrics, any part of the actigraphic recording with involuntary activity caused by sleeping on a rocking crib, or being held by parents. Abnormal data 1. Continuous activity during nighttime and no sleep could be scored (suspect device patterns/artifacts malfunction) 2. Unreasonably high levels of activity and/or light (if available), even during nighttime (suspect device malfunction) 3. Unreasonably low levels of activity during daytime, when the wearer is most active (suspect device malfunction) S20 ANCOLI-ISRAEL ET AL.

These periods should be compared to the actigraphy log (or event markers), and when an actigraph removal is noted, the editor should delete the activity data. 3. Note abnormal or unusual movements in the recording: If an actigraphic recording shows abnormal patterns of activity for any part of the record, it may indicate that the actigraph has malfunctioned and that part of the recording should not be used. In some instances, parents interacting with sleeping children can lead to the appearance of wakefulness during sleep. This may occur when the parent is rocking the child whilethe child is asleep or if the child naps in the car. In this case, an actigraphy log or event marker can be used to document this activity so the scorer can accurately identify movement artifacts. Documented times can be used as a guide.

CONSIDERATIONS FOR RESEARCH

A common issue in research is that recording start and end times may need to be synchronized. In such cases, a universal start time can be set for all participants. This may mean that the device start time may be before, during, or after the participant starts wearing the device. The instructions above should be followed to omit any erroneous data.

CHAPTER 5 Scoring Nighttime Sleep/Wake and Considerations for Daytime Naps

Once the actigraphy recording is reviewed and edited, sleep scoring can begin. In some systems, sleep is automatically scored when the ﬁle is opened. In other software packages, the user must manually score the data by selecting speciﬁc algorithms and settings, and then activating the analysis manually. It is also possible to manually score sleep and wake on an epoch-by-epoch basis in most software programs.

SCORING SLEEP AND WAKE

Algorithm and Parameter Selection If different sleep-scoring algorithms (modes) are available, the appropriate one should be designated ﬁrst (see chapter 2 for an explanation of the different acquisition and display modes). In some software packages, the same sleep scoring algorithm can be used for difference modes of activity collection. Some actigraphy software packages also provide a variety of scoring algorithms for adult and pediatric applications. Check the literature and relevant information to make sure that an appropriate algorithm, validated for the speciﬁc device and population to be studied, is selected and used. SBSM GUIDE TO ACTIGRAPHY MONITORING S21

Sleep analysis parameters then need to be chosen, if applicable. Typically, users will employ the default settings in the scoring software; however, at times, adjustments must be made. These parameters may include activity thresholds for sleep or wake, sleep onset and offset times, the deﬁnition of long sleep or wake episodes, and so forth. These settings need to be conﬁrmed each time the software is used, as some packages revert to default settings each time the software is activated.

Major Sleep Period For most patients, the key variables of interest will be sleep/wake during the major sleep period at night. A key step, then, is to accurately determine when that period of time begins and ends. This period is typically deﬁned as the window of time spent in bed between the bedtime (beginning time) and rising time (end time) as noted on the actigraphy log and with the event marker. In some cases, these will not be available, and the scorer will need to rely instead on information within the activity data itself to estimate. Many software packages automatically mark the major sleep period as a “rest” or “down” interval. While these may be fairly close approximations to the major sleep period in normal sleepers, in many patient groups with sleep disorders, the accuracy of these time windows is less certain and should not be considered as more accurate than the actigraphy log. In some cases, the beginning and end of the major sleep period will be so ambiguous that the recording for that speciﬁc night cannot be used; however, in some cases, the scorer can use the following guidelines to estimate the sleep window:

1. Bedtime is reﬂected as a sharp decrease in activity. This decrease may not be as pro- nounced if an individual is inactive for an extended period before getting into bed (e.g., elderly adults with limited mobility). In such instances, there may actually be a spike of activity as the individual transitions into bed (e.g., in older sedentary patients with low activity levels). The bedtime may also appear as a sharp decline in light levels (if the device has an integrated light sensor) when the patient turns off the lights or places the arm wearing the actigraph under the bedcovers. 2. Rise time is reﬂected as a sharp increase in activity, sometimes accompanied by an increase in light levels if rise time occurred before sunrise.

When determining the beginning time and end time, the following situations should be taken into consideration, if applicable:

1. The documented times may not be accurate, due to wearers’ memory bias, or the time differences between the clock used by the wearer and the time that the actigraph reads internally. Efforts to increase the patient’s investment in accurately maintaining the actigraphy log will reduce the risk of recall bias. 2. The wearer may go to bed late and rise late on weekends and holidays or may do some shift work. These variations make the sleep periods difﬁcult to ascertain on some nights, particularly if the presleep routine is different. S22 ANCOLI-ISRAEL ET AL.

3. The actigraph devices typically do not adjust to changes in time zones or to Daylight Saving Time transitions. This results in a systematic mismatch between the actigraphy log and the device times. The scorer should adjust accordingly.

Major Sleep Period Interval Placement The time period between the beginning time and the end time should then be set as the in-bed time interval. This may also be referred to as the “rest” or “down” interval by some actigraph manufacturers.

Major Sleep Period Intervals and Editing In order to avoid incorrect estimation of sleep/wake parameters, any incomplete in-bed intervals (see below for deﬁnition) should be removed from the summary of the data using the following rules, which are based on expert consensus:

1. In some instances, the device may be set to begin recording after the patient has gone to bed for the night. For the ﬁrst major sleep period interval, if data collection starts 1 hour or less after the habitual or documented bedtime, this can be used as the ﬁrst in-bed interval; if data collection starts more than 1 hour after the habitual or documented bedtime (i.e., an incomplete in-bed interval), this incomplete in-bed interval should be

omitted from the summary analysis. 2. For the last in-bed interval, if the actigraph is removed, or stopped recording for any reason 1 hour or less before the habitual or documented rising time, set it as the last in-bed interval; if the actigraph is removed, or stopped recording for any reason more than 1 hour before the habitual or documented rising time (i.e., an incomplete in-bed interval), this incomplete in-bed interval should be omitted. 3. If the time of an artifact or device removal, or the total time of all artifacts or device removals, lasts 1 hour or longer within 1 in-bed interval, this interval should be omitted.

Sleep/Wake Data Review After all in-bed intervals are correctly set and all removals are appropriately edited, review the software-generated sleep/wake data to assure that all needed major sleep period intervals are included, and all incomplete major sleep period intervals (as deﬁned above) are excluded.

SCORING NAPS AND DAYTIME SLEEP

Information about sleep outside of the major sleep period can be clinically informative, and some software packages have the capability to examine data on napping and daytime sleeping. It is important to consider, however, that the sleep scoring algorithms validated for nighttime sleep estimation are not as well validated for scoring of daytime sleeping and naps. The following methods are recommended to estimate napping if this is relevant to the clinical presentation. SBSM GUIDE TO ACTIGRAPHY MONITORING S23

Recording Naps on the Actigraphy Log The wearer should be directed to document any naps and any actigraph removals during all out- of-bed times (intervals) by marking them on the actigraphy log and/or event marker, particularly if daytime napping or inadvertent dozing is suspected. This will typically be sufﬁcient to capture intentional naps during the recording period.

Out-of-Bed Interval Placement A time period that lies between two major sleep period intervals, or before the ﬁrst or after the last major sleep period, is an out-of-bed interval. This may also be referred to as the “active” or “up” interval by some actigraph manufacturers. In some cases, software will automatically generate statistics for the periods of time outside the major sleep period. In other cases, these periods will need to be demarcated manually, following the same procedures as those used to identify the major sleep period at night. These out-of-bed interval guidelines are particularly relevant for scoring naps.

Out-of-Bed Interval Editing For the time period before the ﬁrst in-bed interval, if data collection started within 3 hours after the previous habitual or documented rising time, set that time period as the ﬁrst out-of-bed

interval; if data collection started 3 hours or more after the previous habitual or documented rising time (i.e., an incomplete out-of-bed time period), delete that incomplete time period to increase the likelihood that all possible naps will be counted during a day. For the time period after the last in-bed interval, if the actigraph is removed, or stopped recording for any reason within about 3 hours (note that there are no hard and fast rules about this time frame as this has not been validated) before the next habitual or documented bedtime, set that time period as the last out-of-bed interval; if the actigraph is removed, or stopped recording for any reason 3 hours or more before the next habitual or documented bedtime (i.e., an incomplete out-of-bed time period), delete that incomplete time period to ensure all possible naps will be counted during a day. If the time of an artifact or a device removal, or the total time of all artifacts or device removals, lasts 3 hours or longer within 1 out-of-bed interval, this interval should be omitted from the nap analysis.

Nap Scoring There are several ways to score napping. If the nap scoring function is provided by the software, use the following steps to score napping:

1. Conﬁrm or change parameters for nap analysis, such as nap onset and offset times and the minimum and maximum length limits of a nap, if desired. 2. Activate the nap analysis function. 3. If the out-of-bed intervals are correctly set, the nap data will be automatically calculated. S24 ANCOLI-ISRAEL ET AL.

4. Review the nap data, making sure all needed out-of-bed intervals are included and all incomplete out-of-bed intervals are excluded.

If nap information is available on the actigraphy log, it may also be used to score napping by employing the following steps:

1. Set the reported nap time period(s) based on reported nap start and end times. 2. Make sure only needed nap intervals are set. 3. Activate the software to calculate nap data, like calculating sleep data for in-bed intervals. 4. Sleep/wake analysis parameters may need to be changed to obtain appropriate nap data. 5. Sometimes the documented nap time may not be accurate, due to wearers’ memory bias, or the time differences between the clock used by the wearer and the time that the actigraph reads internally. If this is the situation, adjust the nap start and end times by referring to activity levels or event markers, if available.

If a nap analysis function is not available, or no information is available about documented naps (diaries, event markers), the software-scored sleep/wake data during out-of-bed intervals could be used to estimate napping. The sleep/wake scoring parameters may need to be adjusted for nap analysis.

CONSIDERATIONS FOR RESEARCH

Exporting Actigraphy Data Scored sleep/wake data may also be exported for research purposes or statistical analysis of aggregated data. Variables of interest should be chosen first and then exported into an appropriate data file format. The scored epoch-by-epoch activity, sleep/wake, light (if available) and interval data may also be exported and analyzed by using other statistical tools, such as SAS®, Stata® or SPSS®. Scored nap data may also be exported for research purposes. In this case, the needed variables should be chosen first, and then exported into an appropriate format. The scored epoch-by-epoch activity, nap, light (if available), and interval data may also be exported and analyzed by using other statistical tools.

CHAPTER 6 Reports and Interpretation

Reports of the results of actigraphy recordings performed for clinical reasons should follow systematic guidelines so information can be compared across time within a patient and so that comparisons can be made among groups of patients and across sites. Sample reports for adults, SBSM GUIDE TO ACTIGRAPHY MONITORING S25 children, and teenagers can be found in Appendices B–E. To optimize the integrity of the data gathered, it is recommended that the parameters and settings used for each recording be documented in the report. This chapter outlines the information required and recommended for inclusion in reports as well as information that is optional, based on the reason for conducting the actigraphy recording.

Justification or Reason Recording Was Conducted Actigraphy reports should open with a statement about the reason the recording was conducted. A statement about the sleep complaint, why actigraphy was needed to help with diagnosis, treatment planning, and monitoring outcomes would also be useful. In many cases, this will inform the duration of the recording. For example, a longer recording time may be required for assessment of circadian rhythm sleep disorders, where it may be critical to capture both weekdays and weekends over several weeks compared to recording for an individual with insomnia for whom treatment outcomes are being assessed that may be reasonably captured with a week of recording. Similarly, older children and adolescents will need a longer recording time to capture weekday-weekend variations due to differences in school days and nonschool days.

Technical Specifications for Device and Settings Each report should specifically state information about the device used and any proprietary software used in the scoring. In addition, the specific algorithm and settings used for sleep scoring should be noted. Specifically, the following details should be provided in each report:

1. Device manufacturer 2. Device model 3. Name and version of software used for scoring 4. Parameters and settings (including default settings within the software) for activity data collection, including epoch length and sampling rate 5. Algorithm and settings (including default settings within the software) for sleep scoring, including the threshold for sleep, if applicable (e.g., high, medium, or low sensitivity), or the name of the algorithm employed (e.g., Cole-Kripke)

Duration of Recording The current requirements for actigraphy recordings for clinical purposes require the recording to be at least 72 hours in length. Both the intended length of the recording and the number of usable days should be included in the report. If some days cannot be analyzed, this should be documented in the report.

Settings of Recording Documentation of the location of the patient during the recording should be included. This will typically be in the individual’s home sleep environment; however, if the individual was in a hospital or institution, this should be documented as well. S26 ANCOLI-ISRAEL ET AL.

Actigraphy Log It should be noted whether the patient was asked to keep a concurrent actigraphy log, that for children and adolescents it is important to note whether the patient or the parent ﬁlled out the actigraphy log. If the actigraphy log was not returned or correctly completed, this too should be noted and any limitations associated with interpreting the results in the absence of a valid actigraphy log should be clearly stated. At times, the actigraphy data will not align with the patient’s actigraphy logs. In this case, documentation about this discrepancy should be included in the report, and the clinician may wish to document the suspected reason for the discrepancy.

Comment on Technical Quality of Recording The technical quality of the recording should be indicated. If there are no technical concerns, a statement such as, “There were no apparent technical issues with the recording. All data were used in sleep scoring and analysis” should be included within the report. If it appears the patient removed the device (or if removals were indicated on the sleep log), this should be documented and the method used to handle device removals should be indicated. That is, if a manual scoring process was used, that process should be noted. If the automatic off-wrist detection was used, that also should be noted.

Standard Variables to Report All actigraphy reports should include a core set of sleep variables for the duration of the recording. In some cases, it may be preferable to report the nightly (and corresponding daily) data as well. The way in which the variables were derived should be speciﬁed carefully, because software varies considerably. It is recommended that the major sleep period be identiﬁed using a patient-recorded actigraphy log or other method (e.g., use of an event marker on the device) rather than relying on automated processes available in some software packages for identifying the major sleep period (as described in chapter 5). The reason for this procedure is that, in patient populations with disrupted sleep (e.g., insomnia), individuals often spend excessive amounts of time in bed, but with considerable movement, resulting in potentially invalid automatically detected rest periods. The following variables should be included in all actigraphy reports:

1. Recorded variables extrapolated from actigraphy log a. Bedtime: Clock time attempted to fall asleep based on actigraphy log or event marker b. Wake time: Clock time of ﬁnal awakening in the morning based on actigraphy log or event marker c. Time in bed (TIB): Duration between reported bedtime and wake time (reported in hours or minutes). For most individuals, this will take place at night; however, for some (e.g., shift workers) it may not. 2. Actigraphy variables a. Sleep onset: Clock time individual fell asleep as determined by the actigraph based on the sleep-scoring algorithm or hand-scoring rules. SBSM GUIDE TO ACTIGRAPHY MONITORING S27

b. Sleep offset: Clock time individual woke up as determined by the actigraph based on the sleep-scoring algorithm or hand-scoring rules. c. Sleep period: Duration between sleep onset and sleep offset (reported in hours or minutes). d. Total sleep time (TST): Duration of sleep during the major sleep period (reported in hours or minutes). e. Total wake time (TWT): Duration of wake during the major sleep period (reported in hours or minutes). The sum of TST and TWT should equal the sleep period. f. Sleep efﬁciency (SE) or sleep percent: There are a number of ways in which SE is computed, and it differs across devices; however, the variable of interest is the proportion of time the patient is asleep out of the total time in bed. Sleep efﬁciency should technically only be calculated when reported variables are available (providing the TIB). Sleep percent, on the other hand, provides the percent of time the individual is asleep during the actigraphically scored in-bed period.

The following variables may be included, but are not required:

1. Sleep onset latency (SOL): Duration between reported bedtime and actigraphically scored sleep onset time. While most software packages will provide an estimate of SOL, this variable requires the presence of reported sleep information from the actigraphy log or event marker. In addition, SOL is less reliable than the required variables listed above for some patient populations, particularly, for patients with insomnia disorders. As a result, this variable should be reported and interpreted with caution. 2. Night waking: Devices and software packages vary widely in how awakenings are defined. Also, the correspondence between these variables and PSG is less consistent than for the required variables listed above. These variables should be reported and interpreted with caution. For research, some investigators use a predetermined number of minutes of wake (e.g., 5 minutes) preceded and followed by predetermined number of minutes of sleep (e.g., 15 minutes). The definition used should be included in the final report. 3. Night waking frequency: number of night wakings; Fragmentation Index. 4. Night waking duration: sum or average duration of night wakings. 5. 24-hour sleep duration:amount of sleep per 24-hour period (reported in hours or minutes). This may be useful in studies of younger children where napping is age-appropriate. 6. Naps (frequency and/or duration): Scoring rules similar to night waking are required to define naps (i.e., predetermined number of minutes asleep preceded and followed by predetermined number of minutes of wake). Nap frequency and nap duration can then be defined, based on these rules. One key advantage of wrist actigraphy is that, because the devices can be worn for 24 hours, information is available about activity patterns during the daytime as well as the nighttime. In some cases, it may be relevant to report the total sleep time, percent time asleep, or other parameters separately for the period of time outside of the major sleep period. 7. Variability across days: Indices of the variability of the sleep parameters may be informative, and in some cases, it may be important to separately report types of days such as weekdays and weekends, school days and nonschool days, to inform interpretation of the findings. This may be particularly relevant in establishing the diagnosis of a circadian S28 ANCOLI-ISRAEL ET AL.

rhythm sleep disorder, where a phase shift is not obvious on days when the sleep schedule is imposed on the patient (e.g., school days) compared to days when the patient is free to choose his or her own sleep schedule (e.g., weekends). 8. Circadian variables: Some devices and software packages provide information on the circadian (i.e., 24-hour) patterns of activity, independent of the sleep/wake scoring. These parameters typically include the rhythm acrophase (timing of activity rhythm peak), amplitude (height of the activity rhythm peak), mesor (midpoint of the activity rhythm from peak to trough), and F-statistic (a measure of robustness of the rhythm). Some devices and software packages provide other variables, which can be reported as well. Statistical software packages can be used to implement algorithms to compute these parameters; for example, an SAS© program is available for research data analysis.17

Report Summary and Interpretation and Clinical Recommendations The report summary should include details of the recording that are relevant for the referral question and any additional information that would inform treatment planning. When actigraphy is used to evaluate response to treatment, the outcome of treatment should be included. For example, the change in sleep efﬁciency with insomnia treatment, or the shift in the sleep schedule after treatment of a circadian rhythm sleep disorder (CRSD), should be reported. When the actigraphy recording does not provide sufﬁcient detail to clarify or in some cases establish a diagnosis or to evaluate treatment outcomes, the report should include a

recommendation for additional diagnostic testing (e.g., extended actigraphy, laboratory PSG). Limitations of the actigraphy recording should also be noted when appropriate. For example, if the study was ordered due to complaints of daytime sleepiness, it should be stated that actigraphy is not a measure of daytime sleepiness per se. However, the ﬁndings from the study suggest that daytime sleepiness may be a result of a certain observed sleep pattern.

Considerations for Research For research purposes, variables beyond those listed above are often required and the data may be needed in the original (rather than aggregated) form. As a result, a summary clinic report often is not needed. Rather than a per-patient report, it is recommended that each study use a set protocol for generating the variable summary, and this document should be included as part of the standard operating procedures or standard operating procedures (SOPs) for each study.

CHAPTER 7 The Future of Wrist Actigraphy

Actigraphy is an evolving technology, and the clinical application of this technology is still in its infancy. As the digital world moves quickly forward, the types of actigraphs and user interfaces SBSM GUIDE TO ACTIGRAPHY MONITORING S29 available, as well as the varied applications for actigraph devices, are also changing. The past decade has seen the emergence of mobile phone applications and activity monitors marketed directly to consumers that emulate actigraphy and allow individuals to monitor their own sleep and activity levels. While these are appealing because of low cost and the user-friendly interfaces, they ﬁrst need to be validated against gold-standard methods before implementation into clinical or research practice. Unfortunately, the limited available studies suggest poor correspondence between these consumer-ready devices and those that have been validated by rigorous research. Actigraphy was traditionally used for scoring of sleep and wakefulness alone. It is now also being used more frequently for the measurement of circadian activity rhythms, and some manufacturers have built automated algorithms into their software to make this an option for clinicians working with patients who have circadian rhythm sleep disorders. For researchers, exported raw data from actigraphy can be modeled using other software (see Marler et al.,17 for example). In the meantime, the use of actigraphs in clinical practice is growing as more clinicians understand the beneﬁts of long-term recordings of sleep/wake patterns in place of (or in addition to) the one-night snapshot offered by laboratory PSG. For sleep clinicians, actigraphy can provide useful informationabout the patient’s sleep over a series of nights. Actigraphy also supplies information about change in sleep over time—in the patient’s home sleep environment—and can inform clinical decision making. In research settings, standardized processes for the use of actigraphy will enhance reliability across studies and lead to easier comparisons across descriptive studies and intervention trials.

REFERENCE LIST

1. Ancoli-Israel S, Cole R, Alessi CA, Chambers M, Moorcroft WH, Pollak C. The role of actigraphy in the study of sleep and circadian rhythms. Sleep 2003;26:342–92. 2. Kushida CA, Chang A, Gadkary C, Guilleminault C, Carrillo O, Dement WC. Comparison of actigraphic, polysomnographic, and subjective asessment of sleep parameters in sleep-disordered patients. Sleep Med 2001;2: 389–396. 3. Morgenthaler T, Alessi C, Friedman L, et al. Practice parameters for the use of actigraphy in the assessment of sleep and sleep disorders: an update for 2007. Sleep 2007;30:519–29. 4. Sadeh A, Sharkey KM, Carskadon MA. Activity-based sleep-wake identiﬁcation: An empirical test of methodological issues. Sleep 1994;17:201–7. 5. Sadeh A, Alster J, Urbach D, Lavie P. Actigraphically based automatic bedtime sleep-wake scoring: Validity and clinical applications. Journal of Ambulatory Monitoring 1989;2(3):209–16. 6. Kripke DF, Mullaney DJ, Messin S, Wyborney VG. Wrist actigraphic measures of sleep and rhythms. Electroen- cephalogr Clin Neurophysiol 1978;44:674–76. 7. Mullaney DJ, Kripke DF, Messin S. Wrist-actigraphic estimation of sleep time. Sleep 1980;3:83–92. 8. Webster JB, Kripke DF, Messin S, Mullaney DJ, Wyborney G. An activity-based sleep monitor system for ambulatory use. Sleep 1982;5:389–99. 9. American Academy of Sleep Medicine, Coding FAQ. c2014. Available from: http://www.aasmnet.org/codingfaq. aspx 10. Montgomery-Downs HE, Insana SP, Bond JA. Movement toward a novel activity monitoring device. Sleep & breathing = Schlaf & Atmung 2012;16:913–7. S30 ANCOLI-ISRAEL ET AL.

11. Meltzer LJ, Avis KT, Valentin J, Ambler D. Direct comparison of Fitbit with polysomnography and actigraphy in children and adolescents. Sleep 2014;37:A312. 12. Gorny SW, Spiro JR. Comparing different methodologies used in wrist actigraphy. Sleep Review 2001;Summer: 40–2. 13. Blackwell T, Ancoli-Israel S, Gehrman PR, Schneider JL, Pedula KL, Stone KL. Actigraphy scoring reliability in the study of osteoporotic fractures. Sleep 2005;28:1599–603. 14. Adkins KW, Goldman SE, Fawkes D, et al. A pilot study of shoulder placement for actigraphy in children. Behavioral Sleep Medicine 2012;10:138–47. 15. Acebo C, Sadeh A, Seifer R, et al. Estimating sleep patterns with activity monitoring in children and adolescents: How many nights are necessary for reliable measures? Sleep 1999;22:95–103. 16. Carney CE, Buysse DJ, Ancoli-Israel S, et al. The consensus sleep diary: Standardizing prospective sleep self- monitoring. Sleep 2012;35:287–302. 17. Marler MR, Martin JL, Gehrman PR, Ancoli-Israel S. The sigmoidally-transformed cosine curve: A mathematical model for circadian rhythms with symmetric non-sinusoidal shapes. Stat Med 2006;25:3893–904.

ADDITIONAL READINGS: REVIEW ARTICLES

Martin, J. L., & Hakim, A. D. (2011). Wrist actigraphy. Chest, 139, 1514–1527. Meltzer, L. J., Montgomery-Downs, H. E., Insana, S. P., & Walsh, C. M. (2012). Use of actigraphy for assessment in pediatric sleep research. Sleep Medicine Reviews, 16, 463–475. Sadeh, A., & Acebo, C. (2002). The role of actigraphy in sleep medicine. Sleep Medicine Reviews, 6, 113–124. Sadeh, A. (2011). The role and validity of actigraphy in sleep medicine: An update. Sleep Medicine Reviews, 15,

259–267.

ADDITIONAL READINGS: VALIDATION STUDIES

Ancoli-Israel, S., Mason, W. J., Clopton, P., Fell, R., Jones, D. W., & Klauber, M. R. (1997). Use of wrist activity for monitoring sleep/wake in demented nursing home patients. Sleep 20, 24–27. Belanger, M. E., Bernier, A., Paquet, J., Simard, V., & Carrier, J. (2013). Validating actigraphy as a measure of sleep for preschool children. Journal of Clinical Sleep Medicine, 9, 701–706. Cellini, N., Buman, M. P., McDevitt, E. A., Ricker, A. A., & Mednick, S. C. (2013). Direct comparison of two actigraphy devices with polysomnographically recorded naps in healthy young adults. Chronobiology International, 30, 691–698. Chae, K. Y., Kripke, D. F., Poceta, J. S., Shadan, F., Jamil, S. M., Cronin, J. W., & Kline, L. E. (2009). Evaluation of immobility time for sleep latency in actigraphy. Sleep Medicine, 10, 621–625. Cole, R. J., Kripke, D. F., Gruen, W., Mullaney, D. J., & Gillin, J. C. (1992). Automatic sleep/wake identiﬁcation from wrist activity. Sleep, 15, 461–469. de Souza, L., Benedito-Silva, A. A., Pires, M. L., Poyares, D., Tuﬁk, S., & Calil, H. M. (2003). Further validation of actigraphy for sleep studies. Sleep, 26, 81–85. Edinger, J. D., Means, M. K., Stechuchak, K. M., & Olsen, M. K. (2004). A pilot study of inexpensivesleep-assessment devices. Behavioral Sleep Medicine, 2, 41–49. Hyde, M., O’Driscoll, D. M., Binette, S., Galang, C., Tan, S. K., Verginis, N., ::: Horne, R. S. (2007). Validation of actigraphy for determining sleep and wake in children with sleep disordered breathing. Journal of Sleep Research, 16, 213–216. Insana, S. P., Gozal, D., & Montgomery-Downs, H. E. (2010). Invalidity of one actigraphy brand for identifying sleep and wake among infants. Sleep Medicine, 11, 191–196. SBSM GUIDE TO ACTIGRAPHY MONITORING S31

Johnson, N. L., Kirchner, H. L., Rosen, C. L., Storfer-Isser, A., Cartar, L. N., Ancoli-Israel, S., ::: Redline, S. (2007). Sleep estimation using wrist actigraphy in adolescents with and without sleep disordered breathing: A comparison of three data modes. Sleep, 30, 899–905. King, M. A., Jaffre, M. O., Morrish, E., Shneerson, J. M., & Smith, I. E. (2005). The validation of a new actigraphy system for the measurement of periodic leg movements in sleep. Sleep Medicine, 6, 507–513. Kripke, D. F., Hahn, E. K., Grizas, A. P., Wadiak, K. H., Loving, R. T., Poceta, J. S., ::: Kline, L. E. (2010). Wrist actigraphic scoring for sleep laboratory patients: Algorithm development. Journal of Sleep Research, 19, 612–619. Kushida, C. A., Chang, A., Gadkary, C., Guilleminault, C., Carrillo, O., & Dement, W. C. (2001). Comparison of actigraphic, polysomnographic, and subjective assessment of sleep parameters in sleep-disordered patients. Sleep Medicine, 2, 389–396. Lichstein, K. L., Stone, K. C., Donaldson, J., Nau, S. D., Soefﬁng, J. P., Murray, D., ::: Aguillard, R. N. (2006). Actigraphy validation with insomnia. Sleep, 29, 232–239. Meltzer, L. J., Walsh, C. M., Traylor, J., & Westin, A. M. (2012). Direct comparison of two new actigraphs and polysomnography in children and adolescents. Sleep, 35, 159–166. Montgomery-Downs, H. E., Crabtree, V. M., & Gozal, D. (2005). Actigraphic recordings in quantiﬁcation of periodic leg movements during sleep in children. Sleep Medicine, 6, 325–332. Montgomery-Downs, H. E., Insana, S. P., & Bond, J. A. (2012). Movement toward a novel activity monitoring device. Sleep and Breathing, 16, 913–917. Paquet, J., Kawinska, A., & Carrier, J. (2007). Wake detection capacity of actigraphy during sleep. Sleep, 30, 1362– 1369. Pollak, C. P., Tryon, W. W., Nagaraja, H., & Dzwonczyk, R. (2001). How accurately does wrist actigraphy identify the states of sleep and wakefulness? Sleep, 24, 957–965. Sforza, E., Johannes, M., & Claudio, B. (2005). The PAM-RL ambulatory device for detection of periodic leg movements: a validation study. Sleep Medicine, 6, 407–413. Spruyt, K., Gozal, D., Dayyat, E., Roman, A., & Molfese, D. L. (2011). Sleep assessments in healthy school-aged children using actigraphy: Concordance with polysomnography. Journal of Sleep Research, 20, 223–232.

Weiss, A. R., Johnson, N. L., Berger, N. A., & Redline, S. (2010). Validity of activity-based devices to estimate sleep. Journal of Clinical Sleep Medicine, 6, 336–342.

APPENDIX A Sleep Watch Instructions and Care

As part of your clinic evaluation, you are being asked to wear a special watch. This watch can tell us about when you are asleep and when you are awake. Please wear the watch on the wrist that you don’t write with (so if you are right-handed, wear the watch on your left wrist) Here are other some important things for you to know.

THE SLEEP WATCH IS NOT A TOY. PLEASE BE VERY CAREFUL WITH IT!

If you are going someplace where the watch might get lost or damaged, please leave the watch at home. Do not try and take the sleep watch apart, as this will ruin it and we won’t be able to use it again. S32 ANCOLI-ISRAEL ET AL.

These watches cost a lot of money to replace (more than if you bought two Apple iPads), so again, please be careful!

The sleep watch is water resistant, but not waterproof.

This means it is okay if the watch gets wet, but if you are showering, bathing, or swimming, you should take the watch off (but don’t forget to put it back on after!).

THERE ARE TWO BUTTONS ON THE SLEEP WATCH

EVENT: You should press the event marker once at the following times:

1. When you ﬁrst try to fall asleep at bedtime (after you turn the light out) 2. When you wake up during the night (if you wake up during the night) 3. When you wake up in the morning to start your day 4. When you try to nap (if you take a nap)

When you press the Event button, you should brieﬂy see an E on the screen. MODE: If you press this button, it will display the date (dat) as MMYY (731 for July 31).

Problems?!? Please call XXX-XXX-XXXX right away!!

APPENDIX B Sample Adult Actigraph Report

PatientName:JaneDoe DOB:06/22/1971 MR#:12345678 DOS:03/18/2014–04/01/2014 Report Interpretation: 04/11/14

Jane Doe is a 42-year-old woman with a history of excessive daytime sleepiness. An actigraphy study was ordered to evaluate Ms. Doe’s sleep patterns in her home environment, including sleep onset time, sleep onset latency, night waking frequency and duration, sleep offset time, and sleep efﬁciency. Ms. Doe was asked to wear an Ambulatory-Monitoring Sleepwatch actigraph for two weeks on her nondominant wrist and keep a concurrent sleep diary. The data were collected in 1-minute epochs, and scored with the Action-W 2.7 software using the Cole-Kripke algorithm. There were no apparent technical issues with the recording. All data were used in sleep scoring and analysis. This report contains an overall summary as well as an actigram, which provides a picture of the full actigraphy study. SBSM GUIDE TO ACTIGRAPHY MONITORING S33

SUMMARY OF OVERALL SLEEP PATTERNS

Ms. Doe pressed a button on her watch each night when she attempted to fall asleep, and these times matched her reported sleep attempt time on the daily diary. Her average reported bedtime was 10:25 p.m. Actigraphy found a sleep onset time of 10:39 p.m. Ms. Doe reported an average wake time of 7:24 a.m., with actigraphy showing his average wake time to be 7:21 a.m. Ms. Doe’s average sleep onset latency was within normal limits at 14.1 minutes, with a sleep onset latency under 15 minutes on 9 out of 14 nights, suggesting a relatively rapid sleep onset at bedtime most nights. Overall, Ms. Doe had an average sleep opportunityof 9.0 hours, with her sleeping an average of 8.1 hours. Ms. Doe’s sleep efﬁciency (actigraphy time sleeping divided by reported time in bed) was normal at 90.6%. On only 1 out of 14 nights was her sleep efﬁciency low (77%), and according to the diary, this was a night where she was up multiple times to care for a sick child. Ms. Doe averaged 0.4 wakings per night (>10 minutes), with a range of 0 to 3 wakings per night. It is notable that on 10/14 nights Ms. Doe had no prolonged night wakings lasting more than 10 minutes.

Weekday Versus Weekend Sleep Ms. Doe’s bedtime was slightly later on weekends than weekdays by report (27 minutes), with actigraphy showing a slightly later weekend sleep onset time (10:25 p.m. vs. 11:03 p.m.) weekdays and weekends. Notably wake time was 95 minutes later both by report and actigraphy on weekends compared to weekdays. Average total sleep time was 58 minutes longer on weekends.

Daytime Naps Actigraphy identiﬁed 10 naps during this study on 7 individual days (with more than one nap occurring on 3 days). On the sleep diary, Ms. Doe identiﬁed that for three of these naps she was sitting quietly with little movement. However, the duration of these three periods and level of activity would suggest that she did indeed fall asleep during these periods. The naps ranged in duration from 20 minutes to 3.4 hours, with an average of 83.8 minutes.

SUMMARY AND RECOMMENDATIONS

Ms. Doe is a 42-year-old woman with a history of excessive daytime sleepiness. Actigraphy per se is not a measure of daytime sleepiness; however, this study conﬁrmed signiﬁcant daytime sleepiness, with frequent naps occurring despite an adequate sleep opportunity, an appropriate nighttime sleep duration, and a lack of prolonged nighttime awakenings. Further evaluation with polysomnography and multiple sleep latency testing may be indicated to clarify the cause and severity of her daytime sleepiness. S34 ANCOLI-ISRAEL ET AL.

APPENDIX C Sample Adult Actigraph Report

PatientName:SarahBrown DOB:03/22/1990 MR#:0113355 DOS:08/18/2014–08/28/2014 Report Interpretation: 09/02/2014

Sarah Brown is a 24-year-old woman with an irregular sleep schedule and daytime sleepiness. An actigraphy study was ordered to evaluate Sarah’s sleep patterns in her home environment, including sleep onset time, sleep onset latency, night waking frequency and duration, sleep offset time, and sleep efﬁciency. Sarah was asked to wear an Ambulatory-Monitoring Sleepwatch actigraph for 10 nights on her nondominant wrist and keep a concurrent sleep diary. The data were collected in 1-minute epochs, and scored with the Action-W 2.7 software using the Cole- Kripke algorithm. There were no apparent technical issues with the recording. All data were used in sleep scoring and analysis. This report contains an overall summary as well as an actigram, which provides a picture of the full actigraphy study.

SUMMARY OF OVERALL SLEEP PATTERNS

Sarah pressed a button on her watch each night when she attempted to fall asleep, and these times matched her reported sleep attempt time on the daily diary. Her average reported bedtime was 4:55 a.m. (range 3:10 a.m. to 7:25 a.m.). Actigraphy found a sleep onset time of 5:35 a.m. (range 3:54 a.m. to 7:41 a.m.). Sarah reported an average wake time of 1:19 p.m. (range 8:45 a.m. to 4:27 p.m.), with actigraphy showing her average wake time to be 1:13 p.m. (8:42 a.m. to 4:12 p.m.). Sarah’s average sleep onset latency was within normal limits at 15 minutes, with a sleep onset latency under 30 minutes on 9/10 nights (the other night was 31 minutes), suggesting a relatively rapid sleep onset at bedtime. Notably on her sleep diary, Sarah reported an average sleep onset latency of 36 minutes (range 15 to 75 minutes). Overall Sarah had an average sleep opportunity of 8.4 hours (range 5.6 to 10.9 hours), with her sleeping an average of 7.2 hours (range 3.6 to 10.4 hours). Sarah’s average sleep efﬁciency (actigraphy time sleeping divided by reported time in bed) was normal at 86% (above 80% on 9/10 nights). Once asleep, Sarah reported few prolonged nighttime awakenings (>10 minutes). By actigraphy Sarah averaged 0.8 wakings per night (>10 minutes, range 0–2, longest waking 46 minutes), with no prolonged wakings on 6/10 nights. Consistently across data points, Sarah’s shortest sleep duration and poorest sleep quality was a result of the one day where she had to wake early for a class, and thus attempted to fall asleep signiﬁcantly earlier than usual (3:00 a.m).

Daytime Naps Actigraphy identiﬁed 5 naps during this study on 5 individual days, consistent with the sleep SBSM GUIDE TO ACTIGRAPHY MONITORING S35 diary. The naps ranged in duration from 14 minutes to 4.8 hours, with Sarah sleeping an average of 1.6 hours.

Total Sleep Duration When considering both naps and “nighttime” sleep duration, Sarah averaged 8.1 hours of sleep per 24 hours (range 5.8 to 11.8 hours).

SUMMARY AND RECOMMENDATIONS

Sarah Brown is a 24-year-old woman with an irregular sleep schedule and daytime sleepiness. The results of this actigraphy study are consistent with a diagnosis of delayed sleep/wake phase disorder, as she is unable to fall asleep before the early hours of the morning, but once asleep has nearly adequate sleep duration and good sleep quality. Although actigraphy is not a measure of sleepiness per se, it is likely that Sarah’s daytime sleepiness is a result of being required to be awake (e.g., for early morning classes) at a time when her biological clock is wanting to be asleep. I discussed at length with Sarah what it means to have a diagnosis of delayed sleep/wake phase disorder and different treatment approaches. Due to the need to be awake for classes that begin as early as 8:00 a.m., Sarah agreed to do a trial of chronotherapy, advancing her sleep-

wake schedule around the clock. Once she achieves the desired sleep schedule, she will use melatonin (0.5 mg taken 4–5 hours before bedtime) and bright light therapy (30–60 minutes) to maintain this schedule 7 nights a week.

APPENDIX D Sample Child Report

Patient Name: Peter Jones DOB: 5/22/06 MR#:098765 DOS:10/3/13–10/17/13 Report Interpretation: 10/25/13

Peter Jones is a 7-year-old boy with asthma and concerns related to insufﬁcient sleep duration and daytime sleepiness. Peter was asked to wear an Ambulatory-Monitoring Sleepwatch actigraph on his nondominant wrist for 2 weeks to assess his sleep patterns in his home environment, including sleep onset and offset times, sleep onset latency, and the frequency and duration of night awakenings. In addition, his parents kept a daily sleep diary to identify the time he attempted to fall asleep and when he woke in the morning. The data were collected in 1-minute epochs, and scored with the Action-W 2.7 software using the Sadeh algorithm. There were no apparent technical issues with the recording. This report contains an overall summary S36 ANCOLI-ISRAEL ET AL. of 12 out of the 14 nights (he did not wear the watch on two nights). Attached is an actigram, giving a picture of the full actigraphy study.

SUMMARY OF OVERALL SLEEP PATTERNS

Peter’s average bedtime (time when he attempted to fall asleep) was 8:19 p.m. Actigraphy found a sleep onset time of 8:33 p.m. Peter reported an average wake time of 6:42 a.m., with actigraphy showing his average wake time to be 6:31 a.m. Peter’s average sleep onset latency was normal at 14 minutes. Overall Peter had an average sleep opportunity of 10.4 hours, with his sleeping an average of 8.5 hours. Peter’s sleep efﬁciency (actigraphy time sleeping divided by reported time in bed) was on the lower end of normal at 82% (normal above 80%). Peter averaged 2.3 wakings per night (>10 minutes), with a range of 1 to 4 wakings per night. However, Peter only had one night with an awakening longer than 25 minutes, and by parent report this was because his siblings woke him early, but he did return to sleep.

Weekday Versus Weekend Sleep Peter’s reported bedtime was later on weekends than weekdays by report (21 minutes), with actigraphy also showing a later sleep onset time on weekends (8:57 p.m.) than weekdays (8:21

p.m.). Wake times by report and actigraphy were similar on weekdays and weekends. Together this resulted in approximately a 30-minute shorter sleep duration on weekends.

Activity During Sleep Restlessness was estimated by activity level during sleep. Peter averaged 52% of the night with no activity. While only 15% of the night included moderate to high levels of activity, 33% of the night included a low level of restlessness that may contribute to his daytime sleepiness.

CLINICAL CORRELATION AND RECOMMENDATIONS

Peter Jones is a 7-year-old boy with asthma, who presents with concerns of insufficient sleep duration and daytime sleepiness. Actigraphy per se is not a measure of daytime sleepiness; however, it does provide an estimate of his sleep patterns for 2 weeks, identifying potential factors that contribute to his sleepiness. In clinic, Peter and his mother reported prolonged sleep onset latency, which was not found with this actigraphy study. In addition, mom reported concerns that Peter was awake for prolonged periods during the night, but since he did not wake her, the frequency and duration were unclear. Again, this actigraphy study did not identify issues with prolonged night awakenings. Since Peter has needed to be awakened several times since the start of the school year, he may benefit from a slightly earlier bedtime on weekdays (e.g., 15 minutes) and a weekend bedtime more consistent with his weekday bedtime. This is especially important as Peter does SBSM GUIDE TO ACTIGRAPHY MONITORING S37 not compensate for a later bedtime on weekends with a later wake time, resulting in a loss of almost an hour of sleep each weekend. Finally, because of his restless sleep, Peter may benefit from having his serum ferritin level checked. Ferritin below 50 ng/mL has been associated with restless sleep in children. Following iron therapy treatment, many children have improved sleep quality. Should Peter be treated for low ferritin, a repeat actigraphy study should be done approximately three months after iron therapy begins.

APPENDIX E Sample Teen Report

PatientName:JohnSmith DOB:9/17/2000 MR#:0224466 DOS:12/6/13–12/12/13 Report Interpretation: 12/19/13

John is a 13-year-old boy with a history of hypersomnolence. An actigraphy study was ordered to evaluate John’s sleep patterns in his home environment, including sleep onset time, sleep onset latency, night waking frequency and duration, sleep offset time, and sleep efﬁciency. John was asked to wear an Ambulatory-Monitoring Sleepwatch actigraph on his nondominant wrist for one week and keep a concurrent sleep diary. The data were collected in 1-minute epochs, and scored with the Action-W 2.7 software using the Sadeh algorithm. There were no apparent technical issues with the recording. All data were used in sleep scoring and analysis. This report contains an overall summary. Attached is the actigram, giving a picture of the full actigraphy study.

SUMMARY OF OVERALL SLEEP PATTERNS

John pressed a button on his watch each night when he attempted to fall asleep, giving him an average bedtime of 10:01 p.m. Actigraphy found a sleep onset time of 11:01 p.m. John reported an average wake time of 8:23 a.m., with actigraphy showing his average wake time to be 8:19 a.m. John’s average sleep onset latency was prolonged at 60 minutes. The last three nights of the study, his sleep onset latency was less than 20 minutes (normal). However, the two nights prior had signiﬁcantly prolonged sleep onset latencies of 156 and 121 minutes. John averaged 2.4 wakings per night (>10 minutes), with a range of 0 to 6 wakings per night. All of his night wakings were less than 30 minutes in duration. However, he had a signiﬁcant number of briefer arousals and restless sleep, with high amplitude activity level noted for over 20% of the night. S38 ANCOLI-ISRAEL ET AL.

Overall, John had an average sleep opportunity of 10.4 hours, with him sleeping an average of only 7.4 hours. John’s sleep efﬁciency (actigraphy time sleeping divided by reported time in bed) was low at 72%.

Weekday Versus Weekend Sleep John’s bedtime was signiﬁcantly later on weekends than weekdays by report (103 minutes), with actigraphy showing a later sleep onset time on weekends versus weekdays (12:50 a.m. vs. 10:17 p.m.). Wake time was approximately 2.5 hours later both by report and actigraphy on weekends compared to weekdays. Notably total sleep time was only 10 minutes longer on weekdays.

Summary John is a 13-year-old boy with a history of hypersomnolence. This actigraphy study identified prolonged sleep onset latency, multiple night wakings, and restless sleep, resulting in a low sleep efficiency. In addition, John had significantly shorter sleep duration during the week, which also contributes to daytime somnolence. Notably, the nights with his longest sleep onset latency followed days with a much later bedtime and wake time. This would suggest that John is experiencing “social jetlag” on weekends, resulting in a circadian shift and contributing to his insufficient sleep. Actigraphy per se is not a measure of daytime sleepiness, but it is likely

that the shortened total sleep time and poor sleep quality found in this study contribute to John’s daytime concerns.

RECOMMENDATIONS

Based on this actigraphy study and clinical correlation, John would beneﬁt from a consistent sleep schedule 7 nights a week, improving his sleep onset latency and sleep duration. In addition, John should have his serum ferritin level checked. Ferritin below 50 ng/mL has been associated with restless sleep in children. AMERICAN ACADEMY OF SLEEP MEDICINE REVIEW PAPER

The Role of Actigraphy in the Study of Sleep and Circadian Rhythms

Sonia Ancoli-Israel PhD,1 Roger Cole PhD,2 Cathy Alessi MD,3 Mark Chambers PhD,4 William Moorcroft PhD,5 Charles P. Pollak MD6

1Department of Psychiatry, University of California, San Diego and Veterans Affairs San Diego Healthcare System, 2Synchrony Applied Health Sci- ences, Del Mar, CA, 92014, 3Geriatric Research, Education and Clinical Center; VA Greater Los Angeles Healthcare System and UCLA School of Medicine, Multicampus Program in Geriatric Medicine and Gerontology, 4Private Practice, Las Vegas, Nevada, 5Colorado State University and Northern Colorado Sleep Consultants, LLC, 6Department of Neurology, The Ohio State University

Citation: Ancoli-Israel S, Cole R, Alessi C et al. The role of actigraphy in the study of sleep and circadian rhythms. American Academy of Sleep Medicine Review Paper. SLEEP 2003;26(3):342-92.

1.0 BACKGROUND Actigraphs today have movement detectors (e.g., accelerometers) and sufficient memory to record for up to several weeks. Movement is sam- ACTIGRAPHY HAS BEEN USED TO STUDY SLEEP/WAKE PAT- pled several times per second and stored for later analysis. Computer TERNS FOR OVER 20 YEARS. The advantage of actigraphy over tra- programs are used to derive levels of activity/inactivity, rhythm param- ditional polysomnography (PSG) is that actigraphy can conveniently eters (such as amplitude or acrophase) and sleep/wake parameters (such record continuously for 24-hours a day for days, weeks or even longer. as total sleep time, percent of time spent asleep, total wake time, percent 1 In 1995, Sadeh et al., under the auspices of the American Sleep Disor- of time spent awake and number of awakenings). ders Association (now called the American Academy of Sleep Medicine, AASM), reviewed the current knowledge about the role of actigraphy in 2.0 OBJECTIVES the evaluation of sleep disorders. They concluded that actigraphy does provide useful information and that it may be a “cost-effective method This paper reviews four major areas in which actigraphy is used for for assessing specific sleep disorders...[but that] methodological issues the measurement of sleep or rhythms. The first area of review covers the have not been systematically addressed in clinical research and prac- more recent papers on the technology and validity of actigraphy. Sadeh tice.” Based on that task force’s report, the AASM Standards of Practice et al. concluded that the validation studies for normal subjects showed Committee concluded that actigraphy was not indicated for routine diag- greater than 90% agreement and were very promising.1 Actigraphs and nosis or for assessment of severity or management of sleep disorders, computer programs using different algorithms to process the data have but might be a useful adjunct for diagnosing insomnia, circadian rhythm been commercially available for quite some time. Actigraphs differ in disorders or excessive sleepiness.2 Since that time, actigraph technology how they detect and record movements and they use different method- has improved, and many more studies have been conducted. Several ologies for computing activity levels. The output of the analysis pro- review papers have concluded that wrist actigraphy can usefully approx- grams has been compared to the results of PSG and sleep diaries. This imate sleep versus wake state during 24 hours and have noted that actig- section reviews the results and evaluates the conclusions of these types raphy has been used for monitoring insomnia, circadian sleep/wake dis- of studies. turbances, and periodic limb movement disorder.3,4 This paper begins The second area of review is of those studies examining actigraphy in where the 1995 paper left off. Under the auspices of the AASM, a new populations with sleep disorders. Actigraphy is being used more often in task force was established to review the current state of the art of this studies of sleep disorders, either as an alternative to PSG, as an addition technology. to partial unattended monitoring devices or for follow-up. This is espe- Actigraphs are devices generally placed on the wrist (although they cially common in patients with complaints of insomnia. can also be placed on the ankle or trunk) to record movement. Collected In addition to gaining information about sleep, data collected over data are downloaded to a computer for display and analysis of activi- long periods can be used to determine activity circadian rhythm cycles. ty/inactivity that in turn can be further analyzed to estimate wake/sleep. Actigraphy is particularly useful for recording rhythms, as it is very dif- The latter technology is based on the observation that there is less move- ficult to record PSG for 24-hours and almost impossible to record for ment during sleep and more movement during wake. As described in more than 24-hours. The use of actigraphy in studies of circadian Ancoli-Israel,5 the first actigraphs were developed in the early 1970’s.6-8 rhythms comprises the third area of review. Kripke and colleagues were some of the first investigators to publish The fourth area of review is those studies in which actigraphy was reliability data on the use of wrist actigraphy for the assessment of used as a treatment outcome measure or to examine the relationship sleep.9-11 Over the years, additional types of actigraphs were developed between sleep/activity patterns and demographic or clinical variables. leading to the digital types used today. Since actigraphy is easier to use, less invasive and substantially less expensive than PSG, actigraphy is often used in lieu of PSG in both clinical trials where it is necessary to determine the effect of a treatment on Disclosure Statement sleep and in studies requiring multiple measurements. Supported by: NIA AG08415, NCI CA85264, the Department of Veterans Affairs VISN-22 Mental Illness Research, Education and Clinical Center (MIRECC), 3.0 METHODS the Research Service of the Veterans Affairs San Diego Healthcare System [to SAI]; NIA AG13885, VA HSRD IIR 01-053-1, the Department of Veterans Affairs As with the first review in 1995, the Standards of Practice Committee Geriatric Research, Education and Clinical Center (GRECC) at the VA Greater of the American Academy of Sleep Medicine commissioned this updat- Los Angeles Healthcare System [to CA]. ed review. A Medline literature search was conducted from the year 1995 to April 2002. Key words for the Medline search included actigraphy, Address correspondence to: Sonia Ancoli-Israel, Ph.D., Department of Psychia- actigraph, actigraphic recording, actimeter, actometer, wrist actigraph, try, VASDHS, 3350 La Jolla Village Drive, San Diego, Ca 92161,USA. Tele- actigraph recording, wrist activity, rest activity, activity, and sleep-wake phone: 858 642-3828; Fax: 858 552-7536; activity, each paired with sleep, sleep disorders and sleep disorders-cir- email: [email protected] cadian. Articles published prior to the original American Academy of SLEEP, Vol. 26, No. 3, 2003 342 Actigraphy Review Paper—Ancoli-Israel et al Sleep Medicine’s (AASM) Practice Parameters for the Use of Actigra- publications not typically found through Medline. In an attempt to phy in the Clinical Assessment of Sleep Disorders1 in 1995 were not include all articles matching the stated criteria, task force members also included in the current update, and only articles written in English were added any articles they discovered through their personal review of the included. literature. All new articles published up to the point of the final draft of A total of 171 articles were identified as potentially relevant based on this manuscript (July 2002) were reviewed. these Medline searches. All of these were obtained in full length and Only papers where actigraphs were used to measure some aspect of examined. Upon review of these articles, approximately 30 additional sleep/wake activity or circadian rhythms were included. Papers that only references were discovered by perling (i.e., checking the reference sec- measured activity (without any reference to sleep) or that made mea- tions for articles otherwise missed). These were references located in surements only in the daytime were excluded. Only papers published in English, in peer-reviewed journals, were included. Case studies and review articles were included in the narrative, but not in the evidence Table 1—Levels of Evidence for Actigraphy tables. No conference abstracts, even if published, were included. Papers Level1 Grade2 Criteria were categorized into four sections: Technology, Sleep Disorders, Circa- 1ABlind, prospective comparison of results obtained by actigraphy dian Rhythms, and Other Clinical Research. to those obtained by a reference standard3 on an appropriate spec- Within each category, task force members were assigned to read each trum of subjects and number of patients. paper, summarize the relevant points for the evidence tables and rate the 2BBlind, prospective comparison of results obtained by actigraphy study according to the evidence levels shown in Table 1. Abbreviations to those obtained by a reference standard3 on a limited spectrum used in the evidence tables are described in Appendix A. of subjects or number of patients.

3CComparison of results obtained by actigraphy to those obtained 4.0 TECHNOLOGY by a reference standard3, but not blind, not prospective or otherwise methodologically limited. The technology section (see Table 2) includes studies that compared functional differences between actigraphic devices, data acquisition 4Ca - Adequate comparison of results obtained by actigraphy to strategies, and software programs. The question asked was, “How well those obtained by a non-standard reference3; or b - Actigraphy not compared to any reference, but actigraph did the devices and the computer programs that process data, assess results demonstrated ability to detect significant difference states of sleep/wake and related phenomena?” between groups or conditions in well-designed trial.

5DActigraphy not adequately compared to any reference, and either 4.1 Data Acquisition a - Actigraph not used in a well-designed trial, or b - Actigraph used in such a trial but did not demonstrate ability Mechanically, the first generation actigraphs were threshold-motion to detect significant difference between groups or conditions. detectors, which were nonlinear and failed to be sensitive enough to detect small movements.12 They also tended to saturate with modest lev- 1 Level refers to level of evidence. 2 Grade refers to grade of recommendation. els of movement. Some of the newer actigraphs detect motion with lin- 3 Reference standards for actigraphic evaluation of sleep and circadian rhythms may ear accelerometers in a single axis or multiple axes.12 Most single axis include, as appropriate, polysomnography, oximetry, melatonin rhythms, core body tem- acceleration devices in use today use .25 to 2-3Hz bandpass filtering perature rhythms, and/or other generally accepted “gold standards,” applied in an accept- before data are stored, essentially eliminating very slow movements of able manner. Non-standard references include such items as sleep logs, spousal reports, other experimental monitors, etc. less than .25 Hz and movements faster than 2-3 Hz. This is consistent with the early recommendations of Redmond and Hegge who noted that voluntary human movement rarely exceeds 3-4 Hz, and that involuntary a. Time above threshold movements such as tremor and shivering exceed 5Hz.13 However, van Someren et al. suggested using 0.5-11Hz bandpass filters that would reduce gravitational artifacts while picking up some of the faster movements that occur in younger subjects.14 After motion is transduced into an analog electrical form, it is digitized and stored. Some aspects of these processes are programmable by the user, such as the length of the epoch over which activity counts are accumulated and stored. Other aspects of the digitizing are built into the b. Zero crossing device. One key component is how the analog signal is digitized: time above threshold, zero crossings, or integration (see Figure 1). The “time above threshold” strategy cumulatively counts the amount of time per epoch that the level of the signal produced in response to motion is above some threshold (commonly 0.1 to 0.2 g). Two potential problems with this strategy are that the degree to which the amplitude is above threshold is ignored and that the acceleration of the movement is not reflected. The “zero-crossing method” counts the number of times per epoch c. Digital integration that the activity signal level crosses zero (or very near zero). Three potential problems with this approach are that the amplitude of the movement is ignored, the acceleration of movements is not registered, and high frequency artifacts may potentially be counted as considerable movement. “Digital integration” involves sampling the accelerometry output sig- Figure 1—Three different methods for deriving activity counts in actigraphy. Panel A, time nal at a high rate, then calculating the area under the curve for each above threshold, derives the amount of time per epoch that the activity is above some epoch. Rectifying the analog signal doubles the amount of data available defined threshold (represented by a dashed line). Panel B, zero crossings, counts the num- for analysis. Digital integration reflects acceleration and amplitude of ber of times the activity reaches zero (represented by the solid baseline) during the epoch. Panel C, digital integration, calculates the area under the curves represented by black shad- movement, however duration and frequency of movements are not ing. (Based on Gorny and Spiro (15)) shown.

SLEEP, Vol. 26, No. 3, 2003 343 Actigraphy Review Paper—Ancoli-Israel et al In direct comparisons of these three methods of deriving activity sensitivity, specificity, and overall accuracy or agreement (true sleep counts, using the same movement input, digital integration was found to epochs + true wake epochs/all epochs) separately.20 Sensitivity for sleep be better for identifying movement amplitude than time above threshold is the proportion of PSG sleep epochs also identified as sleep by actig- and both digital integration and time above threshold were better than raphy. Specificity for sleep is the proportion of non-sleep (wake) epochs zero crossing.15 Some devices simultaneously utilize more than one correctly identified by actigraphy. Sensitivity and specificity for wake- method of acquiring data thus increasing the benefits and reducing the fulness are similarly defined. Agreement is the proportion of PSG deficits of utilizing only one method. epochs correctly identified by actigraphy. Sensitivity, specificity and Many investigators have begun to report actigraphy data simply as agreement therefore assess actigraphic recorders using an established activity counts. However, different devices, different data collection standard. Data collected using this scheme have shown that actigraphy is strategies,12 and different scoring algorithms produce very different more likely to detect sleep (sensitivity) but less reliable at detecting counts for the same activity.15 These differences have made direct com- wake (specificity) (evidence level 1A and 3C).19,21 parisons between laboratories and clinics difficult and contentious. More recently, Pollak and colleagues have suggested different ways to Although relative changes in activity can be meaningful, more direct compute the performance of an actigraph’s ability to detect sleep and comparisons following computer processing of the data (such as wake.21 They use “predicted value for sleep” (PVS) which is the pro- sleep/wake scoring) are more meaningful. portion of actigraphic sleep epochs that are also classified as sleep by PSG, and use “predicted value for wakefulness” (PVW) which is the pro- 4.2 Data Processing portion of actigraphic wake epochs that are also classified as wakefulness by PSG. PVS answers the question, “What percentage of the epochs Actigraphy data today are generally processed by computers, after the that the actigraph scores as sleep are true (PSG) sleep?” This is not the data are downloaded from the actigraphs, rather than by hand or eye. same as sensitivity for sleep, which answers the question, “What per- While most programs are designed to work with a specific device, some centage of true (PSG) sleep epochs are detected by the actigraph?” Pol- are intended to work with data produced by several, if not all, devices. lak and colleagues also report agreement rates but correctly point out Different programs often use different algorithms. However, there are no that this measure is actually not very useful because it confounds PVS published articles comparing the different algorithms. Rather, published with PVW. For example, a high PVS but a poor PVW during the sleep articles present information about the accuracy and/or usefulness of the period would yield a high agreement rate since most of the epochs would outputs of specific computer programs. Before choosing an actigraph to be sleep. This high agreement rate gives a false sense of validity since use, two questions need to be asked and answered satisfactorily about much of wake after sleep onset (WASO) could be mis-scored by actig- such programs and the device(s) they work with: raphy because of the low PVW. In fact, it has generally been shown that 1. Is the output reliable, that is, how well does the same input result in actigraphy is better at detecting sleep (high PVS) than at detecting wake the same output? (PVW) during the sleep period (evidence level 1A and 3C).19,21 Results 2. Is the output valid, that is, how well does the output actually measure of future studies are likely to result in more useful measures if predictive what it is purported to measure? values are reported in addition to sensitivity, specificity and agreement. There are only a few published reports of direct studies of reliability It is worth noting that all of these measures vary with the proportion of between actigraphs and PSG. In general, when tested, actigraphy devices recorded PSG epochs that represent sleep (base rate of sleep). During the have been found to be reliable. Jean-Louis et al. compared new and old sleep period, even malfunctioning recorders that are not being worn or instruments of the same make and model in healthy adults and found no are insensitive to movement will appear to accurately identify sleep if differences when the devices were worn on the same wrist (evidence the low activity counts recorded by them are interpreted by scoring algo- 16 level 3C). In a second study when healthy adults wore two actigraphs, rithms as “sleep”. one on each wrist or two on the same wrist, correlations of activity counts were 0.80 to 0.96; when these data were converted to sleep/wake 4.3 Comparisons to PSG scores the agreement rates between pairs of devices ranged from 93% to 99% (evidence level 1A).17 The correlations between devices for data As PSGs are still considered the gold standard, most studies have collected just at night were between 0.60 and 0.96 and the sleep/wake compared actigraphy to PSG. A technical difficulty in doing these com- scoring agreements again were between 93% and 99%.17 Pollak and col- parisons is accurately time-locking the epochs of the actigraph with leagues had their subjects wear actigraphs from two different manufac- those of the PSG.21 If they gradually drift apart, over time different seg- turers and found that there were differences between them (evidence ments of sleep or wake may be compared with each other, rendering the level 3C).18 results meaningless. A related problem is how to compare the often-used There are many published reports on the validity of actigraphs for 1-minute epoch of actigraphy with the standard 30-second epoch of PSG. measuring sleep/wake. Many of these are reviewed in other sections of In these comparisons, low threshold actigraph algorithms (e.g., defin- this report. When comparing actigraphic to other biological variables, ing wake as occurring even when a small number of activity counts their similarity needs to be taken into account. If the actigraphic and accumulated during the epoch) yielded the best accuracy rate and PVS, other variables are judged to have equal standing, a measure of correla- however, as sleep efficiency diminished, accuracy rate diminished (evi- tion is appropriate, but if one of the variables is invested with definitive dence level 1A).19 Actigraph PVW was best with high threshold algo- importance (“gold standard”), different measures are needed to quantify rithms (e.g., defining wake as occurring when a large number of activi- how close actigraphy approaches the gold standard. For studies of sleep, ty counts, such as 100, accumulated during the epoch) compared to low PSG is considered definitive. Most tests of actigraph validity, therefore, threshold algorithms but at a cost of lower accuracy and PVS.19 One involve comparing actigraphy with PSG. Comparison by correlation pro- computer program, Actigraph Data Analysis Software (ADAS), which vides information about the relative, but not the absolute performance of converts raw actigraph data into information about sleep/wake, has been actigraphy. For example, a high correlation on total sleep time would shown to be valid when compared to PSG (evidence level 2B),22 even mean that individuals who sleep longer by PSG criteria also sleep longer when raw data were derived from different devices and in patients with by actigraph criteria, and vice versa, but this could be true even if actig- insomnia (evidence level 3C).16 raphy systematically over-estimated total sleep time. Therefore, it is pos- Other studies have also shown that actigraphy was highly correlated sible for the correlation between actigraph and PSG sleep results to be with PSG for differentiating sleep from wake (evidence level 2B and 3C high, even when the epoch-by-epoch agreement is relatively low. Corre- respectively),22,23 with reported correlations for total sleep time (TST) lations provide an incomplete picture, especially if most of the data being 0.97.22 Comparisons showed 91%-93% overall agreement in come from the sleep period.19 adults (age 20-30 years) (evidence level 2B),24 and 91.4%-96.5% An alternate approach is Tryon’s method of calculating and reporting

SLEEP, Vol. 26, No. 3, 2003 344 Actigraphy Review Paper—Ancoli-Israel et al minute-by-minute agreement rates in adolescents (age 10-16 years) and logs and actigraphy declined over time apparently because the parents adults (age 20-30 years) (evidence level 1A).17 In healthy adults, actig- increasingly tended to omit items from the logs. Thus, for determining if raphy was valid for assessing sleep durations and sleep/wake activity, a child’s night awakenings decline during treatment, actigraphy appears but less reliable for more specific measures such as sleep offset or sleep to be more accurate (evidence level 4C-a).35 efficiency (evidence level 3C).25 There was also no first night effect in In a study of children and adolescents, Sadeh found that a minimum healthy adults (evidence level 3C).16 In nursing home populations, of seven nights of actigraphy were needed to get five nights of useful Ancoli-Israel et al. reported correlations between actigraphy and PSG data for sleep onset and number of minutes of wake (evidence level 4C- for TST of 0.81 - 0.91 and for percent sleep of 0.61 - 0.78 (evidence level a).33 In another study of children and teenagers, Acebo et al. reported 3C).26 good agreement between observations and actigraphy for sleep onset, There are discrepant reports about the validity of actigraphy for other number of minutes awake, and sleep efficiency (evidence level 4C-a).36 sleep variables. One study of four adults found good correlations for Acebo et al. also found that, in children, more than seven nights of data sleep onset latencies, wake time after sleep onset, sleep efficiency, and collection were needed to get useful data for sleep efficiency, sleep peri- total sleep time (evidence level 3C),27 while two other studies of healthy od and number of minutes of sleep.36 adults found poorer relationships for sleep onset latency and wake time after sleep onset with quiet wake being frequently misidentified as sleep 4.5 Comparisons to MSLT (evidence level 2B and 3C respectively).22,23 Early actigraph validation research found that the correlation between actigraphy and PSG for In a study of the effects of diphenhydramine vs. placebo on daytime sleep onset latency was only 0.53 when sleep onset was defined as the sleepiness, Roehrs et al. showed that MSLT was more sensitive than 37 first minute of actigraph-estimated sleep, but jumped to 0.94 when sleep actigraphy to sleep loss (evidence level 2B). Yet, their data showed that onset was defined as the beginning of the first period containing 20 min- actigraphy during the day reflected prior sleep loss with more epochs of utes of actigraph-identified sleep with no more than one minute of wake inactivity, suggesting more daytime sleep. Since actigraphy is not intervening.28 However, subsequent research has continued to use the restricted to use in a laboratory as is the MSLT, the authors concluded first-minute definition. This may account for some of the observed error that actigraphy during the day may yield a more accurate index of the in actigraphic scoring not only of sleep onset latency, but also of vari- effects of sleepiness. ables that depend upon it, such as sleep efficiency and wake time after sleep onset. A study of healthy adults on a shiftwork schedule found poor 4.6 Comparisons to EMG relationships with sleep efficiency and less reliability for determining Since the actigraph records movements, placement on the foot can be 25 sleep offset compared to PSG (evidence level 3C). Yet others found used to record movements that are most typical of periodic limb move- that actigraphy overestimated sleep efficiency and total sleep time in ment disorder (PLMD). In a study of PLMD, Kazenwadel et al. found 19 patients with sleep disorders (evidence level 1A). To summarize, when actigraphy recorded on the foot to be comparable to surface EMG ante- compared to PSG, actigraphy was found to be valid and reliable for rior tibialis measurements during PSG (evidence level 1A).38 The corre- detecting sleep in normal, healthy adult populations but less reliable for lations between leg kicks determined by activity counts and tibialis 22 detecting sleep as sleep became more disturbed (evidence level 2B). EMG measurements remained high both on and off medication. The authors concluded that measurement of PLMD by actigraphy was possi- 4.4 Comparisons to Observations, Sleep Logs, and Diaries ble if 0.5-second epochs were used. However, considerable manual Actigraphy has also been compared to both direct observations of adjustment and editing of computerized data were necessary to avoid sleep and to sleep logs and diaries. In a study of the effects of circadian underestimating the number of leg kicks. These results were in dis- rhythms entrainment (entrained vs. free-running) on sleep by Lockley et agreement with those of Sforza et al., who also compared actigraphy al., sleep logs and actigraphy yielded similar data for sleep timing, sleep recorded from the foot to PSG-recorded anterior tibialis EMG (evidence 39 duration, sleep onset and sleep offset but not for sleep latency, number level 5D-b). However, the Sforza et al. study had a very small sample and duration of night awakenings or number of naps (evidence level size for a validity study, and only two out of 35 patients had PLMD. (For 1A).29 Nurses’ observations of sleep in psychiatric patients were similar additional discussion on the use of actigraphy in PLMS see also section to actigraph data but sleep logs kept by patients in the morning were not 5.5, Restless Legs Syndrome/Periodic Limb Movement Disorder) found to be satisfactory (evidence level 4C-a).30 Observations of nursing home residents by research staff yielded a PVS of 87% and PVW of 90% 4.7 Actigraph Placement 26 when compared to actigraphy (evidence level 3C). Two studies found no difference between data collected from acti- Monk et al. compared both actigraphs and sleep diaries to PSG during graphs placed on different locations (e.g., dominant wrist, non dominant 31 space flight (evidence level 3C). Predicted values of actigraphy were wrist, ankle, or trunk) (evidence level 3C and 1A respectively).16,17 How- clearly superior to those of diaries for sleep onset and offset, sleep dura- ever, in a series of two studies by Middelkoop et al., other results were tion, and sleep efficiency. The authors concluded that in general, actig- found. In one study, wrist placement was shown to detect more move- raphy is a simple, efficient means of evaluating sleep in situations, such ments than ankle placement which in turn detected more movements 31 as space, when PSG is too cumbersome for routine use. Dijk et al. mea- than trunk placement in the first study (evidence level 4C-a).40 In a sec- sured wrist activity continuously in five astronauts during 10 to 16 days ond study of healthy adults, wrist placement was again superior to both of space flight, and performed sleep PSG on four of those days (evidence ankle and trunk placement, however, dominant wrist placement was bet- 32 level 3C). They found that actigraphically estimated sleep duration was ter than all other placements at detecting wake (evidence level 4C-b}.41 significantly longer on PSG-recording nights than non-PSG nights. They Violani et al. found that the right wrist recorded more activity than the concluded that astronauts probably adhered more closely to their sched- left wrist both early and late in the sleep period but no differences were uled bedtime when their work duties included PSG sleep recording. found between ankle placements (evidence level 4C-a).42 Middelkoop Actigraphy appears to be useful in other populations where PSG and colleagues concluded that more studies which compare different might be difficult to obtain, such as in nursing home patients as men- placements of actigraphy concomitant with PSG recordings were need- 26 tioned above, or in infants and young children. In general, Sadeh con- ed.41 cluded that with infants, actigraphy should be paired with parental sleep logs for screening infant sleep problems, although actigraphy appeared 4.8 Artifacts in Actigraphic Recordings to be a more consistent measure than parents’ sleep logs of the child’s sleep/wake (evidence levels 4C-a).33,34 The agreement rates between the There are some potential artifact problems when using actigraphy for

SLEEP, Vol. 26, No. 3, 2003 345 Actigraphy Review Paper—Ancoli-Israel et al sleep/wake determinations. For example, artifacts can come from non- that may alter a patient’s typical sleep patterns. compliance (not wearing the recorder), from breathing movements, from The convenience of using actigraphy to evaluate disordered sleep, postural blocking of arm movements, or from externally imposed move- however, must be weighed against its reliability and validity as comment from riding in vehicles.21 Many investigators routinely have vol- pared to the traditional gold standard for sleep assessment, polysomnog- unteers who wear actigraphs keep concomitant logs of sleep times and raphy. In addition, a determination of actigraphy’s potential usefulness actigraph removal, and use these data to help with artifact rejection.43,44 must take into consideration how it compares to alternative methods that may be equally or less expensive, such as self-report. Although it might 4.9 Summary be expected that the objective and unbiased nature of data produced by the actigraph would necessarily be more accurate than those yielded by Recent research has refined the ability of actigraphs to study subjective assessment techniques such as sleep logs, this assumption sleep/wake. Although there are still some technical differences in the must be confirmed empirically. mechanical aspects of how actigraphs accumulate data on movements, how these data are processed, and the nature of the algorithms used to 5.1 Insomnia process these data, both the actigraphs themselves and the algorithms that process the data from actigraphs have improved since the last task- Chambers, in an analysis of previously published data,49 found that force report published in 1995.1 For example, a method using dichoto- when total sleep time estimates from actigraphs and sleep logs were mous indices of activity has been developed to compare activity in-bed compared to polysomnography for a group of insomnia patients, there to out-of-bed that, among other things, might be useful for studying cir- was no significant difference in the mean absolute error for the two tech- cadian phase shifts.45 However, no head-to-head comparisons have been niques (evidence level 4C-b).50 Moreover, sleep log estimates of total made between actigraphs and no conclusions can be drawn about which sleep time had a significantly higher correlation with PSG than did those method is more valid vs. PSG. from actigraphy, suggesting that for insomnia patients as a group, sleep Actigraphy is useful in populations where PSG would be difficult to log error, at least with respect to the estimation of total sleep time, is record, such as in demented patients,26 and in astronauts in space.31 more systematic and predictable for sleep logs than for actigraphy. How- There does not seem to be a first night effect with non-sleep disordered ever, this same analysis did reveal a substantial within-subjects, or night- patients (evidence level 3C),16 which is of particular benefit when only to-night correlation (r = 0.81) between actigraph and PSG total sleep one night of recording is possible. If more nights are needed, the acti- time. Such a finding indicates that those factors contributing to actigraph graph also has the advantage of being easy to record for multiple nights. error for a given patient (e.g., periodic leg movements, minimal activity There may be advantages of using combined data from actigraphy and a during extended periods of nocturnal wakefulness) tend to be consistent subjective questionnaire with sleep-disordered patients, especially if from night to night. Therefore, at least for insomnia patients, actigraphy they are excessively somnolent (evidence level 1A).19 For studies of may be useful in the assessment of sleep variability or in the measure- rhythms or for studies where time of lights out is important, an actigraph ment of treatment effects. that also records light exposure would be beneficial.5,46 A number of recent studies have employed actigraphy in the evalua- Yet, regardless of the technology, research studies must demonstrate tion of sleep of the patient with insomnia; however, few of these studies that actigraphy serves the needs of sleep/wake researchers and clini- have validated the actigraphy findings with PSG data. Guilleminault and cians. Given the expanded use of actigraphy, the time has arrived for colleagues in a study of non-drug treatments for insomnia, evaluated standards to be established, similar to those developed for polysomnog- potential subjects using a sleep questionnaire, one week of sleep diaries, raphy in 1968 by Rechtshaffen and Kales.47 Such standards might and four days of actigraph monitoring (evidence level 4C-b).51 These include device standards (e.g., digital integration is best) and/or counts subjects also underwent one night of laboratory polysomnography, but defined with standardized units of measurements (e.g., g-force units) so simultaneous recording with an actigraph was not performed, so direct data from different machines and algorithms could be compared and validation of the actigraphy with PSG findings was not possible. How- comparisons could be made to other acceleration measures. In addition, ever, baseline data from this study did show that actigraphy consistently bench-test minimal standards for computer programs used with actigra- produced higher estimates of total sleep time and the number of awak- phy need to be developed. enings and lower estimates of sleep onset latency than those yielded by Ultimately, field tests are needed to determine what actigraphy is sleep logs. The differences between actigraph and sleep log data were capable of doing and how well it can do it. Data showing that actigraphy attenuated somewhat in the post-treatment recordings, with greater treat- is reliable and valid are necessary as are data demonstrating the best pro- ment-related improvement seen in sleep log variables than in actigraph cedures for getting the best measurements for sleep/wake evaluations variables. (e.g., best location on the body to place the device, how to analyze raw Wilson et al., in a study of insomnia patients with musculoskeletal data) and evaluating potential problems. Published reports using actigra- pain, found a similar discrepancy between the sleep estimates reported phy must contain complete reporting of sensitivity, specificity, scoring on sleep logs and those determined by the actigraph, with a much larger algorithm, and filters, as well as reliability, validity, ruggedness, and arti- disagreement in the number of awakenings during the night (evidence fact rejection for the device and computer program used.18 On the other level 5D-b).52 These researchers found relatively low correlations among hand, technical standards may not be as important as simply demon- patients (r = 0.34 to 0.42) between the two measures for estimates of strating validity and reliability for determination of sleep/wake status of total sleep time and number of awakenings, and no significant correla- all actigraphic measurements with any given apparatus and scoring algo- tion for sleep efficiency. Consistent with the analysis of Chambers (evi- rithm. dence level 4C-b),50 the highest correlation found in this study was for night-to-night actigraphic estimates of total sleep time. However, acti- 5.0 ACTIGRAPHIC ASSESSMENT OF CLINICAL SLEEP DISORDERS graph sleep variables consistently failed to produce significant correlations with clinical assessment measures such as pain severity estimates One possible application of actigraphy in sleep medicine has been the or scores on the Pittsburgh Sleep Quality Index. Two other studies also diagnosis and assessment of clinical sleep disorders (see Table 3). Com- failed to find correlations between actigraph sleep variables and global pared to traditional polysomnography, the actigraph is relatively unob- subjective reports of well-being, sleep behaviors, and health-related trusive and can record for multiple days and nights. This may be useful symptoms (evidence level for both 5D-b).53,54 in the assessment of insomnia patients, whose sleep has been shown to Wicklow and Espie examined the relationship between cognitive 48 be quite variable from night to night. Moreover, actigraphy makes intrusions prior to sleep onset and sleep-related variables as measured by home recording more accessible, permitting the evaluation of patients in actigraphy and sleep logs (evidence level 4C-a).55 Their findings indi- their natural sleeping environment and eliminating laboratory effects cated that the presence of certain categories of intrusive thoughts was

SLEEP, Vol. 26, No. 3, 2003 346 Actigraphy Review Paper—Ancoli-Israel et al associated with longer sleep-onset latencies, but only as measured by melatonin administered in the evening for two weeks restored sleep con- actigraphy, not sleep logs. These researchers also found that actigraphic tinuity in a child with a germ cell tumor involving the pineal region.65 TST was greater and sleep latency was less than that indicated by sleep Some studies have used actigraphy to examine developmental differ- logs. However, there were significant correlations (r = .419 for sleep ences in sleep patterns. Sadeh et al. recorded the sleep of school-age latency, r = .526 for TST, both p < .001) between the two methods for children for 4 to 5 nights and found that older subjects had delayed these variables. sleep-onset times, shorter sleep periods, and shorter sleep times than younger subjects (evidence level 4C-b).66 They also found that increased 5.2 Insomnia Secondary to Circadian Rhythm Disturbance reported daytime sleepiness was associated with greater age and shorter sleep periods, as measured by the actigraphy. Aronen et al. demonstrat- Other researchers have utilized actigraphy to assess insomnia com- ed that actigraphically measured TST was negatively correlated with plaints secondary to a circadian rhythm disturbance. Kerkhof and van teacher-reported behavioral symptoms in young children (evidence level Vianen divided a group of chronic insomniacs into early- and late-sleep 4C-b).67 Kramer et al., although not specifically studying children, did phase groups, based on oral body temperature data, and found greater show apparent developmental differences in sleep between young and nocturnal motor activity, as measured by actigraph, in the early phase elderly adult subjects (evidence level 4C-b).68 They found the elderly 56 group (evidence level 4C-b). This finding was consistent with the sub- subjects, whose average age was 65 years, to have less variability for jective assessments of these subjects, who reported spending more time time in bed, advanced sleep phase, and more nocturnal awakenings than awake during the night than the late phase group. Several studies have the younger subjects (mean age = 20.6 years). used actigraphy to assist in the diagnosis of delayed sleep phase syn- 44,45,57-61 drome (DSPS) and to assess effects of DSPS treatments. DSPS 5.4 Sleep-Related Breathing Disorders is characterized by a consistent pattern of late sleep onset and offset, often making diagnosis by PSG impractical. Although sleep logs are Several study protocols have attempted to detect the presence of often used to diagnose DSPS, actigraphy can potentially offer objective obstructive sleep apnea from actigraphic data. This work has generally evidence about rest-activity patterns that either corroborates logs or calls relied on the fact that compared to normal sleepers, apnea patients have them into question. Dagan et al. used 4-7 days of actigraphy at home in more fragmented sleep and that this fragmentation is manifested in body conjunction with a comprehensive clinical assessment to diagnose movements that can be detected by the actigraph. Middelkoop et al. dozens of subjects with DSPS, but they offered no independent evidence found that the average duration of periods with no movement (i.e., no of the validity of this method (evidence level 5D-a).57 Similarly, Quinto activity), as measured by the actigraph, differed significantly among et al. reported that actigraphy “confirmed” sleep logs in a case of three subject groups of varying apnea severity (evidence level 4C-b).69 DSPS.59 Minors et al. reported that there were significant differences in No other actigraph or sleep log variable correlated significantly with the wrist activity patterns that distinguished people with DSPS from normal apnea index. However, the proportion of variance accounted for by this sleepers (evidence level 4C-b).45 Nagtegaal et al., in two separate stud- variable was small (11%), and the sensitivity of this measure to detect ies, provided evidence that wrist activity patterns in DSPS were consis- subjects with an apnea index greater than 5 was 5% while the specifici- tent with a “gold-standard” biological marker of circadian rhythm dis- ty was 100%. turbance. In the first study (a case report), they found that the late time Drinnan et al. attempted to determine whether the specific placement of day of low activity corresponded with a late period of high melatonin of the actigraph might affect its accuracy in the identification of arousals secretion in a case of DSPS.60 In the second study (a randomized, place- associated with sleep-disordered breathing (evidence level 5D-b).70 bo-controlled trial), they found that actigraphy detected a 38-minute Their data revealed that a left tibia placement resulted in the most favor- advance in sleep-onset time with melatonin treatment, paralleling able relationship between actigraph-measured movement and EEG advances in dim-light melatonin onset. Sleep log variables, however, arousals, with placement on the right tibia, left ankle, and left wrist far- failed to detect this phase shift (evidence level 1A).61 Cole et al. also pro- ing somewhat worse. Still, none of the placements yielded statistically vided objective evidence that actigraphy can detect circadian rhythm significant correlations with EEG arousals, and none were adequate in disturbance in DSPS (evidence level 2B).44 They reported that the circa- predicting the degree of sleep disordered breathing present. As with the dian phase of melatonin secretion was significantly delayed, compared Middelkoop et al. study (evidence level 4C-b),69 however, the relatively to normal, historical controls, in 45 DSPS volunteers whose delayed low severity of sleep apnea among patients in this study (mean apnea- sleep was identified both by actigraphy and sleep logs. Additional infor- hypopnea index or AHI = 18.9) may have limited the power of the tech- mation on the use of actigraphy in circadian rhythms can be found in nique to differentiate between groups. section 6.0. Kushida et al. compared PSG, actigraphy and subjective reports in a study of 100 sleep clinic patients, the majority of whom had a diagnosis 5.3 Disturbed Sleep in Children of obstructive sleep apnea syndrome or upper airway resistance syndrome (evidence level 2B).19 Consistent with previous studies, they Actigraphy has also been used to assess disturbed sleep in children. found that the actigraph was considerably better at detecting sleep than Franck et al. compared sleep, recorded by actigraph, of HIV-infected detecting wakefulness, with a sensitivity of 98% for sleep detection and children to that of normal controls and confirmed that sleep-related com- a specificity of 48% for wake detection using a high-threshold algo- plaints, as reported by parents, were greater in the patient group (evi- rithm. This algorithm compared PSG and actigraphic data in 30-second 62 dence level 4C-b). Actigraph estimates of sleep efficiency, WASO, and epochs, modifying the activity counts during the epoch by the level of number of awakenings were significantly different between the two activity in the surrounding 2-minute time period. When compared to groups, while differences in TST and SOL failed to reach significance. PSG, the actigraph was much more prone to overestimate total sleep time Within the patient group, the only significant correlation between acti- and sleep efficiency than was subjective patient report. However, the graph data and subjective reports was for night waking. Another study actigraph’s estimates of number of nocturnal awakenings did not differ examined autistic children with and without parent-reported sleep prob- significantly from PSG data while self-report did, suggesting actigraphy lems and found an earlier sleep offset time for the sleep problem group was more accurate than the patient’s subjective reports. but no other significant differences in actigraphically measured sleep Elbaz et al. reported on an inventive use of actigraphy in the diagno- 63 variables (evidence level 5D-b). Sadeh et al. using actigraphy, report- sis of sleep-disordered breathing that combines the actigraph with what ed that newborns slept twice as much during night time hours than dur- they termed “simplified polysomnography,” consisting of airflow, tho- ing the day and that later gestational age was correlated with an racic and abdominal movements, and pulse oximetry (evidence level 64 increased percentage of quiet sleep time (evidence level 4C-b). In a 3C).71 They reasoned that the addition of actigraphy could improve the case report, Etzioni et al., using wrist actigraphy, found that 3mg of estimation of the RDI from that of simplified polysomnography alone by

SLEEP, Vol. 26, No. 3, 2003 347 Actigraphy Review Paper—Ancoli-Israel et al supplying a more precise value for TST than the traditionally used TIB. many cases over a substantial period of time. However, these case stud- However, the actual increase in correlation with RDI estimates from tra- ies did not compare the actigraphy data to PSG, nor did they indicate ditional polysomnography was somewhat modest, from r = .94 for sim- whether the actigraph alone was sufficient to diagnose the conditions. plified polysomnography alone to r = .976 for simplified polysomnography plus actigraphy. The authors did find that specificity and negative 5.7 Summary predictive value were substantially improved with use of the actigraph, but only for severe OSAS (RDI > 30). Recent literature suggested that actigraphy might have some value in the assessment of sleep disorders. For insomnia, actigraphy may be most 5.5 Restless Legs Syndrome/Periodic Limb Movement Disorder valuable in assessing treatment effects or night-to-night variations in subjects’ sleep. It has been demonstrated that actigraphy has the ability One of the more natural applications of actigraphy has been in the to detect sleep phase alterations associated with circadian rhythm distur- identification and assessment of periodic leg movement disorder bances. Additionally, actigraphy is capable of distinguishing moderate to (PLMD). Sforza et al. conducted a study of 35 patients with varying severe sleep apnea patients from normal controls, due to its greater sen- diagnoses to determine if actigraphy could reliably detect leg move- sitivity, compared to sleep logs, in detecting brief arousals from sleep. ments during sleep (evidence level 5D-b).39 Subjects were simultane- For patients with restless legs and periodic leg movements, the diagnos- ously recorded using PSG and actigraphy placed on the upper part of the tic value of actigraphy is limited by its tendency to underestimate the fre- right foot. PSG-recorded EMG tibialis activity was visually scored and quency of leg movements during sleep. However, it does show some classified in 8 levels based on duration and amplitude. Actigraphic data promise in the assessment of treatment-related improvement. were collected in 5-second epochs and compared directly to EMG activ- Later-generation scoring algorithms have demonstrated greater accu- ity of similar-length epochs. The results of this analysis revealed that racy than earlier versions in the detection of sleep and wake, improving although there was a high correlation between the two collection meth- the actigraph’s ability to detect sleep latency, nocturnal awakenings, and ods, actigraphy substantially underestimated the number of movements total sleep time, variables important in evaluating an insomnia com- yielded by the EMG. However, this failure may be attributable to the low plaint. The actigraph appears to have a particular advantage over alter- sensitivity of the actigraph used, which is only able to detect accelera- native assessment methods such as sleep logs in the measurement of tions greater than 0.1 g. In contrast, Pollak used actigraphs that were sen- awakenings during the night, as many of these awakenings appear to go sitive to 0.033 and 0.024 g, respectively (evidence level 4C).18 For undetected by patients and subjects completing sleep diaries. The abili- Sforza, agreement was greater for the activity events with greater dura- ty of actigraphy for detecting activity also holds some promise in the tion and amplitude. Because of the actigraph’s failure to detect many identification of other disorders characterized by frequent movements events of lesser duration or amplitude, this study’s authors concluded such as obstructive sleep apnea and periodic limb movement disorder. that the device “cannot be regarded as a good method to estimate motor Perhaps of greatest importance is the actigraph’s ability to measure activity during sleep” (p. 158). However, other authors did note that the night-to-night changes in sleep patterns within a given individual, a actigraph had adequate night-to-night reliability and suggested that it function that has great value for assessing treatment effects and other might be useful in the assessment of treatment effects in patients with factors that affect the consistency of a patient’s sleep. This, combined PLMD or restless legs syndrome (RLS) (evidence level 4 C-b and 5 D- with its relative economy in assessing sleep-wake patterns over extend- b, respectively).50,52 ed periods of time, suggests a potentially important role for the actigraph Two recent studies have employed actigraphy in the evaluation of in longitudinal research and clinical studies in which long-term changes treatment efficacy for RLS.72,73 Trenkwalder et al. in a placebo-con- in sleep patterns are of particular interest. trolled crossover design, studied the effects of L-dopa therapy for idio- The actigraph’s limitations, however, continue to restrict its value as a pathic and uremic restless legs syndrome, using both PSG and actigra- stand-alone diagnostic device. Recent research has reasserted the find- phy at baseline and at the end of each treatment period (evidence level ings from previous studies that the accuracy of the actigraph to detect 1A).72 Their data revealed that treatment resulted in a significant reduc- sleep and wakefulness declines as sleep efficiency is decreased, a prob- tion of leg movements, measured by both PSG and actigraph, for both lem particularly relevant to insomnia and other sleep disorders. There patient types. Both PSG and actigraphic data also showed that this are indications that for the simple estimation of total sleep time, insom- improvement was limited to the first 4 hours of recording time. Parallel nia patients’ subjective estimates outperform the actigraph (evidence to these objective indices, subjective measures such as sleep diaries and level 4 C-b).50 Moreover, although the actigraph may be able to distin- quality of life ratings showed similar improvement in response to treat- guish patients with a particular sleep disorder (e.g., obstructive sleep ment. Furthermore, because actigraphy was continued for two addition- apnea, periodic limb movement disorder) from normal controls, there is al nights after the PSG study, the authors were able to confirm the sta- virtually no evidence to date that the actigraph can distinguish between bility of this treatment effect. different sleep disorders. Until such evidence becomes available, the Collado-Seidel et al. conducted a similar study of L-dopa and slow- actigraph’s function in the assessment and diagnosis of clinical sleep dis- release L-dopa efficacy, but without the use of polysomnography (evi- orders is likely to be restricted to the role of an adjunct to clinical histo- dence level 4C-b).73 Like Trenkwalder et al.,72 these researchers found ry, sleep diary data, and PSG findings or to examine treatment effects significant treatment-related effects for most actigraphic variables, and follow-up. including movements per hour and number of movement episodes. Only the change in the time without movements in the first half of the night 6.0 CIRCADIAN RHYTHMS failed to reach significance. Subjective improvements were also seen with patients reporting increases in sleep quality and overall well-being Activity is a standard marker of circadian rhythms in studies of non- and decreases in number of awakenings, time awake, and reports of day- human mammals. This section examines the use of wrist activity in the time fatigue. measurement of circadian rhythms in humans. In the studies reviewed here, wrist actigraphy was used in a number of different ways relevant 5.6 Other Sleep Disorders to human rhythms. Methodologies included characterizing spontaneous rhythms in adults, children, infants and the elderly, exploring the rela- Various case reports have employed actigraphy in the assessment of tionships between activity rhythms and the light-dark cycle, helping to other sleep disorders, including fatal familial insomnia,74 non-24-hour identify sleep or rhythm disturbances induced by change of schedule, sleep-wake syndrome,75,76 REM sleep behavior disorder,77 and posttrau- measuring improvement in disturbed rhythms after experimental inter- matic delayed sleep phase syndrome.59 In each of these reports, the acti- vention, helping to diagnose circadian rhythm sleep disorders, charac- graph provided data relevant to the patients’ sleep/wake patterns, in terizing rhythm abnormalities that accompany dementia or psychiatric

SLEEP, Vol. 26, No. 3, 2003 348 Actigraphy Review Paper—Ancoli-Israel et al disturbance, and investigating the role of motor activity in cardiovascu- lag,92,97 illness,91 or experimental manipulations of the sleep/wake lar rhythms. cycle,87,98 or by showing98 that sleep is improved by treatment that nor- malizes rhythms.93,99-101 For example, both Dawson et al.93 and Yoon et 6.1 Actigraphy for the Study of Circadian Rhythms al.102 found that actigraphic indicators of sleep improved during the day following simulated night shift work in volunteers treated with bright Several studies have demonstrated that human wrist activity often light or melatonin, but not in those treated with placebo (evidence levels shows a robust circadian pattern. Pollak et al. showed that the circadian 4C-b). Actigraphy was also used to infer when shiftworking nurses slept period of the actigraph-defined sleep/wake rhythm accurately predicted and, along with shifts of the melatonin rhythm, demonstrated that a sub- the period of the PSG-defined sleep/wake rhythm, measured simultane- group of nurses was able to successfully adapt to rapid changes in work- 21 ously (evidence level 3C). shift (evidence levels 4C-b).94,95 Actigraphs have measured circadian rhythms under circumstances Although three shift work studies reported negative results,103-105 where it would not be practical to record with polysomnography. For enough well-designed, well-controlled studies showed significant dis- example, Binkley measured wrist activity in one woman continuously turbance of actigraph-defined sleep after shift work or other circadian 78 for an entire year. She found well-defined, entrained circadian rest- disturbance to make a convincing case that actigraphy can be useful for activity cycles, changes in sleep length synchronized with the menstrual detecting such disturbances (evidence levels 3C to 4C-b).87,91-95,97,98,106 cycle, and annual phase changes she attributed to daylight saving time. Wirz-Justice and colleagues presented several case reports in which acti- 6.2 Actigraphy Algorithms for Computing Circadian Rhythms graphs were worn daily by demented or psychiatric patients for extended periods of up to 1.5 years.79-84 The long-duration recordings allowed As noted earlier, the study of rest-activity rhythms has a long history. the authors to produce plots that graphically revealed striking changes in Prolonged actigraphic recordings lasting for multiple circadian periods circadian rhythms over time. These included severe, apparently medica- can therefore give valuable chronobiological information, even if no tion-induced disruption of the rest activity cycle, the gradual consolida- attempt is made to convert the rest-activity rhythm to the sleep-wake tion of the cycle upon change of medication, circadian effects of rhythm. To extract this information, the raw activity values are analyzed imposed therapeutic rest/activity schedules, and the gradual decline of directly. The most popular method has been cosinor analysis,43,89-92,107,108 circadian organization over time. Siegmund et al. measured seven-day in which a cosine curve with a period at or near 24 hours is fit to the data wrist activity rhythms in inhabitants of Papua New Guinea, living in a by the least-squares method. The parameters that are of interest are traditional culture without electric lights (evidence level 4C).85 They acrophase (time of peak activity), amplitude (peak-to-nadir difference) found that rest-activity rhythms were synchronized with the light-dark and mesor (mean) of the fitted curve (see Fig 2). A “five-parameter cycle, and that time of arising was more consistent than bedtime. Acti- extended cosinor analysis” has also been used to provide a better fit to graphic recordings in infants showed that circadian activity rhythms activity data, which typically deviate from the shape of a cosine arose from ultradian antecedents. Periods of inactivity presumably curve.109,110 The five model parameters are circadian minimum, ampli- encompassing sleep were shorter (9-12 hours/day) than typically found tude, acrophase, alpha (width of the rhythm) and beta (steepness of fit- in European society. In another actigraphic study of infants, “sleep” dif- ted curve, which can approximate a square wave if beta is high). F-statis- ferences were not explained by differences in temperament (evidence tics for goodness-of-fit derived from this model have been used in stud- level 5D-b).86 Dijk et al. measuring wrist activity during 10 to 16 days ies of nursing home patients to detect a significant strengthening effect of space flight, visually identified imposed advances in the sleep-wake of light treatment on circadian activity rhythms (evidence level 4C-b),110 schedule, and noted that time of arising was more regular than bedtime and a significant weakness of the activity rhythm relative to rhythms of (evidence level 3C).32 behavioral agitation and environmental light exposure (evidence level In the studies cited above, and many others,87-91 circadian rhythm 4C-b).109 Van Someren and colleagues found that two analyses that make results were computed from actigraphic sleep/wake predictions. Usual- no a prioi assumptions about the waveform of activity data, autocorrela- ly, the phase marker was sleep onset time or sleep offset time. Similarly, tion and interdaily stability, showed significant strengthening of activity Binkley marked circadian phase with visually identified “activity onset” rhythms in demented patients in response to light therapy, while simple and “activity offset,” defined by threshold criteria similar to those used cosinor analysis (and other analyses that assume a fixed waveform) in some sleep/wake prediction algorithms (evidence level 4C-b).92 showed no significant effect in the same data sets (evidence level 4C- Another way actigraphic sleep/wake predictions (or similar “activity b).111 Autocorrelation is the correlation between activity values at spe- level” scores) have been used to yield circadian results is by showing cific time lags of interest. High autocorrelation at or near 24-hours indi- that actigraph-identified sleep is disturbed when people attempt to sleep cates a robust circadian rhythm. Interdaily stability is a measure of the out of phase with their endogenous rhythm, as in shift work,93-96 jet strength of coupling of a rhythm to environmental zeitgebers, based on the chi-square periodogram, which, in turn, is based on wave-form eduction (see below). Some investigators have computed the “circadian quo- Period tient” (amplitude/mesor) to characterize the strength of the circadian rhythm (more robust rhythms have a higher amplitude, but people who move more vigorously may also have higher amplitude; the circadian Amplitude quotient expresses amplitude relative to mesor, providing a normalized value that allows comparison between individuals).91,112 A similar nor- Activity Mesor malized variable that does not rely on the assumption of a cosine fit is

relative amplitude (based on the difference between the most active 10- hour interval and the least active 5-hour interval of the day).111 Additional circadian outcome measures that have been computed directly from raw activity data include the ratio of nighttime activity to daytime activity or total activity (evidence level 2B and 4C respective- 06 12 18 00 06 12 18 ly),91,99 standard deviation of sleep onset time,113 intradaily variability Acrophase Time (based on the changes in activity level from hour-to-hour),100,111,114-116 89-91,117 Figure 2—Circadian rhythm representation in which a cosine curve with a period at or near various types of spectral analysis, and waveform eduction. Wave- 24 hours is fit to the data by the least-squares method. The parameters that are of interest form eduction is carried out by calculating an “average waveform” for are acrophase (time of peak activity), amplitude (peak-to-nadir difference) and mesor some period. For example, if a period of 24.0 hours is chosen, succes- (mean) of the fitted curve.

SLEEP, Vol. 26, No. 3, 2003 349 Actigraphy Review Paper—Ancoli-Israel et al sive activity levels at similar times of the 24.0-hour day are averaged. 6.4 Actigraphy and Circadian Rhythm Sleep Disorders Sleep-wake consolidation is the extent to which continuous bouts of sleep and wakefulness are clustered into periods that are circadian (last There is good evidence that actigraphy can detect circadian phase for several hours). Waveform eduction can be quantified by a measure- delays in people with DSPS, corresponding to delays in melatonin ment related to the lengths of the “stair treads and risers” in cumulative rhythms (level 1A to 2B) e.g., (44,60,61). This evidence is summarized plots of sleep vs. wakefulness.21 above in the section 5.2. Insomnia Secondary to Circadian Rhythm Dis- turbance. There are also case reports in which actigraphy identified sys- 6.3 Actigraphy versus Other Methodologies for Determining Circadian tematic delays of the rest-activity cycle in non-24-hour sleep-wake syn- 75,76 Rhythms drome. Several studies compared circadian outcomes derived from wrist 6.5 Actigraphy and Circadian Rhythms in Aging and Dementia actigraphy to those derived from other measurement methods considered reference standards. Pollak’s finding that actigraphic sleep/wake predic- Actigraphy has been used to explore circadian rhythms in aging and tions have the same circadian period as PSG sleep/wake scores suggest- dementia. Three studies found that the overall activity level declined 108,122,123 ed that actigraphy could provide valid measurements of entrained with age (evidence level 4C-b to 5D-a), as did the amplitude of 116,124 sleep/wake rhythms (evidence level 3C).21 Youngstedt et al. provided the circadian rest-activity cycle (evidence levels 4C-b). Fragmenta- good evidence that the phases of actigraph-identified bedtime, wake-up tion (hour-to-hour variability) of the rest-activity rhythm was found in 101,116 time, mid-sleep time and acrophase were all significantly correlated with healthy elderly and could be reduced in elderly men by aerobic 101 the acrophase of urinary 6-sulphatoxymelatonin secretion in entrained training (evidence levels 4C-b). young and elderly adults living at home (evidence level 1A).43 Cole et Mishima found increased overall activity and nighttime activity in 125 al. found similar home results in volunteers with delayed sleep phase Alzheimer’s disease (evidence level 4C-b). Friedman et al. showed syndrome (DSPS) (evidence level 2B).44 Middleton et al. found that the that actigraphic measures of circadian activity (including amplitude, phase and period of actigraph-derived sleep onset time, wake-up time acrophase and mesor) did not correlate significantly with behavioral dis- and fitted cosine were generally consistent with those of both 6-sulpha- turbance in patients with Alzheimer’s disease (AD) (evidence level 4C- 126 toxymelatonin and “demasked” core body temperature in men undergo- b). Martin et al. found little evidence of sundowning (increased agita- 109 ing experimental manipulations in constant dim light (both evidence lev- tion around sunset) (evidence level 4C-b). Actigraphic rest-activity els 2B).89,90 However, these studies also showed that activity rhythms rhythms in demented patients were stabilized by increased illumination 100 were less stable than melatonin or temperature rhythms, and could be if vision was intact (evidence level 4C). readily masked by voluntary behavior. Carskadon et al., studying adolescents, found correlations ranging from .39 to .82 between actigraph- 6.6 Actigraphy and Cardiovascular Rhythms identified sleep onset time at home and salivary dim light melatonin Another group of studies used sleep and wake activity to help distin- 118,119 onset time (evidence level 2B to 3C). However, in one study the guish “dippers” (people whose blood pressure decreases normally at correlation dropped to a non-significant level under an imposed light- night) from “non-dippers” (blood pressure that remains the same or rises 118 119 dark cycle, and in the other it was not significant on weekends. during sleep). Although sleep wake activity was not necessarily corre- Heikkilä et al. found that in children suffering a severe medical disorder, lated with blood pressure or heart rate (evidence level 5D-b),127 two the circadian rhythm of wrist activity could be grossly disturbed despite studies found that dippers have lower nocturnal activity than nondippers normal rhythms of melatonin, temperature and cortisol (evidence level (both at evidence level 4C).128,129 Daytime activity levels were also cor- 120 5D). Guilleminault et al. found that consolidated wakefulness, visual- related with the nocturnal dip in BP (evidence level 4C).129 By using ly scored from wrist activity, only developed in infants after they estab- actigraphy to define sleep and wake periods, a calcium-channel blocker 121 lished a circadian rectal temperature rhythm (evidence level 3C). Bla- was found to have therapeutic effects in hypertensives that differed grove et al. provided very strong evidence that actigraph-identified sleep according to the time of day it was administered (evidence level 4C- was influenced directly by the central circadian pacemaker (presumably b).130 A third study found that defining “night” as the actigraph-identi- the suprachiasmatic nuclei, or SCN of the hypothalamus) (evidence level fied sleep period yielded very different blood pressure results than did 87 3C). They measured wrist activity during a forced desynchrony proto- defining “night” by fixed clock time criteria (evidence level 5D-a).131 col in which volunteers lived on a 27-hour day (9 hours in bed, 18 hours up and active), so the pacemaker free-ran at a period closer to 24 hours, 6.7 Actigraphy and Circadian Rhythms in Psychiatry and sleep was attempted at various circadian phases. Despite the imposed rest-activity rhythm, actigraph-identified total sleep time was Actigraphy has also been used to examine circadian rhythms in psy- lower when sleep was attempted at an unfavorable phase of the circadi- chiatry. Wirz-Justice and colleagues found severe disturbance of rest- an cycle. This strongly suggested that the influence of the circadian activity rhythms in one bipolar individual82 and several schizophrenic clock on actigraph-identified sleep could not be entirely masked by a individuals recorded for extended periods (evidence level 5D-a).79,81,83 socially-dictated rest/activity schedule. On the other hand, the masking Similarly, Martin et al. found that both rest-activity rhythms and acti- effect of the imposed schedules was substantial, and this was likely to be graph-identified sleep were often seriously disturbed in 28 older true of actigraph studies in general. Because of their susceptibility to schizophrenics (evidence level 4C-b).132 The magnitude of disturbance masking, wrist activity rhythms alone cannot be used as pure markers of was associated with the degree of neuropsychological impairment. Neu- SCN circadian output, even though they contain an SCN signal. Actig- roleptic-induced akathisia was associated with increased motor activity raphy has been used to “demask” the circadian temperature rhythm, by in schizophrenic patients, at least at certain times of the day (evidence removing the effects of activity and inactivity (evidence level 5D-b).98 level 4C-a).133 Two studies reported that specific depressive syndromes Another line of evidence that actigraphy can accurately characterize were characterized by distinctive circadian activity rhythms.107,134 Glod the circadian sleep/wake rhythm is that the rhythm of actigraph-inferred et al. found blunted circadian amplitude but normal phase of wrist activ- sleep/wake generally agrees with that of sleep/wake reported on sleep ity in children with Seasonal Affective Disorder, compared to healthy logs (evidence level 3C).29 This raises the question of whether sleep logs controls (evidence level 4C-b).107 Lemke et al. found that depressed are just as good as actigraphs for circadian measurements. Two studies adult inpatients displayed significantly greater motor activity in the suggest a possible advantage of actigraphy, that is, that it may identify morning than the evening (evidence level 4C-b).134 naps that volunteers do not report on their sleep logs (both evidence level These studies, taken together, provide preliminary evidence that actig- 4C-a).88,103 However, actigraphy may also identify naps when none exist. raphy may prove useful for characterizing and monitoring the circadian

SLEEP, Vol. 26, No. 3, 2003 350 Actigraphy Review Paper—Ancoli-Israel et al rhythm disturbances that often accompany psychiatric disorders. and did not use hypnotics (evidence level 4C-b).138 Although there were no differences between groups when the 24-hour period was considered 6.8 Summary as a whole, post-hoc comparisons in the early morning hours indicated that subjects using bedtime medications became active, as measured by In summary, actigraphy has been used successfully in a variety of actigraphy, about 1.5 hours earlier in the morning than controls. human circadian studies. Wrist activity appears to be a valid marker of entrained PSG sleep phase, and a strong correlate of entrained endoge- 7.2 Actigraphy in Studies of Healthy Adults nous circadian phase. Under non-entrained conditions, wrist activity rhythms may become dissociated from the endogenous rhythm of the Several studies involving normal individuals under differing testing SCN pacemaker; however, actigraphy still appears to be useful for iden- situations have used actigraphy as a measure of sleep/wake or circadian tifying disturbed sleep caused by disruption of circadian rhythms, and rhythms. Duka et al. measured wrist movement in a placebo-controlled improved sleep caused by treatments that improve rhythms. There is evi- study of the effects of a beta-carboline benzodiazepine receptor antago- dence that the circadian phase of wrist activity covaries with the phase nist on night sleep pattern in healthy male volunteers (evidence level 5D- of melatonin secretion in DSPS, supporting the use of actigraphy in a).139 Compared to placebo, the benzodiazepine receptor antagonist helping to diagnose this condition. Actigraphy may also be useful in cir- induced activation as measured by actigraphy (i.e., frequency of move- cadian characterization of non-sleep disorders, such as schizophrenia ment and intensity of movement). French et al. used actigraphy to mea- and hypertension. A variety of methods for analyzing circadian aspects sure sleep patterns in military aircraft crew members undergoing simu- of activity data show promise. It would be useful to formally compare lated, long duration bomber missions (evidence level 4C-b).140 They these to arrive at standard methodology. found shorter sleep duration and greater wrist activity during sleep periods during the first mission, with evidence of improvement in sleep in 7.0 OTHER CLINICAL RESEARCH subsequent missions. In a study of the effects on sleep from caffeinated beverages in healthy volunteers, Hindmarch et al. found a dose-depen- Actigraphy has been used as a measure of sleep/wake activity or cir- dent negative effect (of caffeine) on total sleep time as estimated by cadian rhythms in a broad range of clinical studies. These studies vary wrist actigraphy (evidence level 4C-b).141 In another study, Hindmarch considerably with respect to the specific actigraphy variables of interest, et al. found that promethazine (a sedating antihistamine) caused a sig- the methodology used and the types of individuals studied. Unfortunate- nificant increase in percent sleep, as estimated by wrist actigraphy, dur- ly, many of these studies do not report adequate detailed information on ing the daytime and across the study period compared with different the technical aspects of the actigraphy devices used, and few studies doses of fexofenadine, loratadine and placebo (evidence level 4C-b).142 attempt validity testing on the use of actigraphy in the particular popu- Jean-Louis et al. analyzed actigraphy data in a large sample (N=273) lation or setting studied. of community-dwelling residents of San Diego who had been identified by random telephone survey (evidence level 4C-b).143 In this cross-sec- 7.1 Actigraphy in Sleep Intervention Trials and Comparative Studies of tional study, they found significant differences between men and Sleep/Activity women, and between Caucasian and minority subjects, in sleep variables Much of the work using actigraphy as a measure of sleep disorders is estimated by wrist actigraphy. In a second smaller community-based reviewed earlier in this paper (see section 5.0). This section focuses on sample (n=32), Jean-Louis et al. used wrist actigraphy in healthy volun- the use of actigraphy as an outcome measure in other sleep intervention teers and again found significant gender differences, with women hav- 122 trials and in comparative studies of sleep or activity. ing a better sleep profile than men (evidence level 4C-b). In a placebo-controlled clinical trial of controlled-release melatonin Mendlowicz et al. performed an observational study using wrist actig- treatment for insomnia in older people (mean age 76 years), Garfinkel et raphy in community dwelling volunteers. In regression analysis they al. reported that melatonin administration resulted in greater sleep effi- found several significant predictors of depressed mood, including the ciency and shorter wake after sleep onset, both estimated by wrist actig- following variables estimated by actigraphy: daytime activity level, raphy (evidence level 4C-b).135 Friedman et al. used multiple modali- sleep onset latency, wake after sleep onset, total sleep time and total time 144 ties, including actigraphy, to measure sleep outcomes in a trial compar- in bed (evidence level 5D-a). Moorcroft et al. used nocturnal actigra- ing the effects of sleep restriction and sleep hygiene treatments on the phy to estimate sleep and wake periods and time of final awakening in sleep of older adults (aged 55 years or older) with insomnia (evidence people who reported the ability to self-awaken at a self-predetermined level 2B).136 The main study outcomes found few between-group differ- time without external means and found that they were in fact able to do ences in treatment efficacy. However, in a sub-sample of 16 subjects so (with a 95% confidence interval of 4.1-10.7 minutes) (evidence level 145 who had simultaneous wrist actigraphy and polysomnography for 3 5D-b). nights, wrist actigraphy estimation significantly correlated with Pankhurst et al., studying the influence of bed partners on nighttime polysomnographic estimation of total sleep time (r = .96), sleep effi- wrist activity in community dwelling adults living in the United King- ciency (r = .63), sleep latency (r = .72) and wake after sleep onset (r = dom, found that subjects sleeping with bed partners had a greater num- .68). In this study, wrist actigraph variables correlated more highly than ber of movements than subjects who slept alone, and movements sleep log data with polysomnography results. decreased during the temporary absence of the usual bed partner (evi- Maus et al. performed actigraphy in a study of circulating cate- dence level 4C-b).146 In a similar but larger sample in the UK, Reyner et cholamines and aqueous flow in the eyes of normal subjects and in those al. found a significant decline with age in average movement as mea- with severe obstructive sleep apnea (evidence level 4C-b).137 Sleep sured by wrist actigraphy, with men having more nighttime movement apnea subjects, who were untreated on the night of testing, had a signif- than women (evidence level 4C-b).147 The authors compared sleep log icantly higher nighttime activity index as measured by actigraphy reports of time of sleep onset with actigraphy estimation of sleep onset, (p<.001) and lower sleep efficiency (p<.001) compared to healthy con- and found that the time difference between the two methods was small; trols. In addition, there were significant differences in activity index and however, the actual time difference was not reported in the manuscript. sleep efficiency in controls who were kept awake during the night versus those allowed to sleep. 7.3 Actigraphy in Studies of Cancer-related Fatigue Pollak et al. studied a small group of community-dwelling elderly Actigraphy has also been used in descriptive studies of cancer-related people who frequently used bedtime medications (including benzodi- fatigue. In an observational study of breast cancer patients, Berger et al. azepines, minor analgesics, antihistamines and antidepressants) and found that greater reported cancer-related fatigue was significantly asso- compared them to elderly controls who did not have sleeping difficulty ciated with a higher number of nighttime awakenings, lower amplitude

SLEEP, Vol. 26, No. 3, 2003 351 Actigraphy Review Paper—Ancoli-Israel et al and lower peak activity as measured by wrist actigraphy (evidence level improvements were consistent with results of actigraphy.156 In another 4C-b).148 In an observational study of cancer patients undergoing radia- sample of 22 men and women undergoing CABG, Redeker et al. found tion therapy for bony metastases, Miaskowski et al. did not find a sig- that activity levels and strength of circadian rhythms increased over days nificant association between wrist actigraphy estimation of nighttime 2 to 5 post-operatively, with a longer time for recovery of activity in sleep and self-ratings of quality of sleep and feeling rested (evidence older adults.155 Finally, in another sample of 33 men and women admit- level 4C-b).149 Using wrist actigraphy to measure circadian rhythms, ted to the hospital for acute myocardial infarction or unstable angina, Mormont et al. found that cancer patients with marked activity rhythms Redeker et al. found that previous severity of heart disease was the had better quality of life, reported less fatigue and had longer survival strongest predictor of lower sleep efficiency and longer duration of compared to those with rhythm alteration (evidence level 4C-b).150 In awakenings during hospitalization.157 multivariate analysis, rest-activity rhythm remained a significant predictor of one-year survival. 7.6 Actigraphy in Studies of Older Adults In a series of articles, Berger et al. reported findings from wrist actigraphy performed in women with breast cancer undergoing several cycles Actigraphy is particularly useful in studies involving older adults, of chemotherapy (both evidence levels 4C-b).148,151 They found that both in the community and in the nursing home. In addition to the com- fatigue ratings were higher during chemotherapy, and negatively corre- munity-based studies described above, which included healthy older lated with activity as estimated by wrist actigraphy. Subjects with lower people in their sample, studies specifically targeting the elderly have circadian measures (specifically lower peak activity) had greater fatigue. used actigraphy as outcome measures. Pollak et al. used wrist actigraphy in a descriptive study of 44 pairs of older people (aged 65 years or older) 7.4 Actigraphy in Studies of Psychiatric Patients with disruptive nocturnal behaviors such as complaining and calling for help, and their principal caregiver (evidence level 4C-b).158 Twenty-two Actigraphy has been used to investigate movement and sleep distur- of the elders met criteria for dementia. Both the older person and their bance in psychiatric patients (circadian activity disturbance is discussed caregiver wore wrist actigraphs. Activity level was less similar during above in section 6.0. Circadian Rhythms). Dursun et al. conducted a the daytime between the older person and their caregiver, as compared descriptive study of wrist actigraphy estimation of sleep in outpatients to nighttime. In addition, actigraphy suggested that at night it was the with schizophrenia prescribed risperidone compared to those on “typi- elders that initiated the elder-caregiver interaction, thus disturbing the cal” antipsychotics, and to normal controls (evidence level 5D-a).152 sleep of the caregiver. They found a greater degree of nighttime wrist movement (i.e., higher Sleep and circadian rhythm variables deduced from actigraph record- movement index) in patients on a typical antipsychotic compared to ings have been used as outcome measures in multiple studies of nursing those on risperidone. Friedman et al. compared wrist actigraphy data home residents, a population that had been understudied and in which with measures of behavioral problems in a sample of patients with PSG is particularly difficult. Ancoli-Israel et al. found significant sleep Alzheimer’s disease (AD) who were participating in a larger longitudi- disruption (with frequent nighttime awakening and frequent daytime nal study (evidence level 4C-b).126 They found that greater behavioral sleeping, based on actigraphy) in a sample of 25 nursing home residents disturbance was correlated with lower actigraphically estimated sleep (evidence level 4C-b).159 In another study, Ancoli-Israel et al. compared efficiency (r=-.35, p<.05) and greater wake after sleep onset (r=.43, nursing home residents with severe dementia to those with moderate, p<.01). mild or no dementia, and found that the severely demented group had Lemke et al. used wrist actigraphy to estimate mean activity levels in lower activity mesor, lower amplitude and were more phase delayed than psychiatric unit inpatients with major depressive disorder (evidence those with moderate, mild or no dementia (evidence level 4C-b).112 level 4C-b).153 They found that subjects whose Pittsburgh Sleep Quality Hourly profiles of sleep and wakefulness in this group suggested that the Index indicated poor sleep had higher mean nighttime motor activity severely demented residents had more sleepiness during the day and levels that those who reported good sleep. In addition, subjects with night and residents with moderate or mild dementia had more wakeful- fewer depressive symptoms had lower mean nighttime motor activity ness during the night (evidence level 4C-b).160 Actigraphy with light levels than those with greater depressive symptomatology. exposure was also used to study the relationship between sleep and light exposure with results suggesting that higher light levels predicted fewer 7.5 Actigraphy in Studies of Adults with Other Specific Medical Conditions nighttime awakenings and later activity acrophase (evidence level 4C- b).161 Additionally, when these same patients were treated with 10 days Actigraphy has been used in a variety of clinical studies involving of bright light therapy, although there was no improvement in sleep at adults with other specific medical conditions. Baker et al. compared night, morning bright light delayed the peak of the activity rhythm (i.e., wrist actigraphy findings between menopausal women and controls, and acrophase), increased the mean activity level (i.e., increased the mesor) found that menopausal women had more arousals and greater sleep dis- and improved activity rhythmicity (evidence level 4C-b).110 ruption. In a study of sleep disturbance in cirrhosis, Cordoba et al. found Alessi et al. used wrist actigraphy estimation of sleep as an outcome that compared to normal controls, cirrhosis patients had decreased motor variable in a controlled clinical trial of physical activity in nursing home activity, more fragmentation of sleep and dampened rhythms, as mea- residents (evidence level 4C-b).162 They found no significant improve- 154 sured by actigraphy (evidence level 4C-b). ment in sleep associated with improved physical function. Likewise, in Redeker et al. reported a series of studies using actigraphy in adults an observational study of incontinent nursing home residents, Alessi et undergoing coronary artery bypass graft surgery (CABG), and in adults al. found no significant differences in nighttime sleep variables between hospitalized for cardiac conditions (all with evidence level 4C-b). In one subjects on psychotropic medications and subjects not on these medica- study, 25 women (mean age 63.7 years) undergoing CABG had wrist tions (evidence level 5D-b).163 However, in a controlled trial of a com- actigraphy applied after admission to the open-heart recovery bined physical activity and environmental intervention in nursing home room/intensive care unit, and wore the actigraphs continuously through- residents, Alessi et al. found a higher percent sleep at night estimated by out their hospital stays. Findings from the first postoperative week after wrist actigraphy in the intervention group compared to controls (evi- CABG, indicated that, after controlling for preoperative functional sta- dence level 4C-b).164 tus, there was a relationship between both recovery from surgery and Cruise et al. performed an observational study in nursing home resi- length of stay with the rhythmic and linear patterns of activity. Positive dents to study the nighttime environment and incontinence care practices linear trends in circadian activity periods were related to better func- in nursing home residents (evidence level 4C-b).165 They found that 42% 155 tioning and shorter length of stay. When wrist actigraphy was repeat- of nighttime waking episodes identified by wrist actigraphy were asso- ed up to four times over the 6 months following CABG, sleep consoli- ciated with noise, light or incontinence care. Ouslander et al. found that dated and daytime sleep decreased and subjects’ perceived sleep

SLEEP, Vol. 26, No. 3, 2003 352 Actigraphy Review Paper—Ancoli-Israel et al nighttime urinary incontinence was not related to sleep disruption (evi- from surgery and length of stay. dence level 5D-b).166 Actigraphy has been used extensively in studies involving older people, particularly in the nursing home setting. These studies have demon- 7.7 Actigraphy in Studies of Children strated significant sleep disruption among nursing home residents, and sleep and circadian rhythm disturbances have been shown to be more Actigraphy has been increasingly used in children, particularly in severe among residents with more severe dementia. In addition, actigra- studies involving children with behavioral, psychiatric or neurological phy has been used to measure treatment effects in sleep intervention illness. Corkum et al. compared wrist actigraphy estimates of sleep in research in the nursing home setting, where other measures, such as children with attention deficit/hyperactivity disorder (ADHD) to normal PSG, would be extremely problematic to perform. controls and found no statistically significant differences in total sleep Finally, there is a growing literature using actigraphy in children. For duration, sleep onset and number of nighttime awakenings by wrist example, actigraphy has been used to demonstrate differences in sleep 167 actigraphy (evidence level 5D-b). Glod et al. used waist placement of between abused children and those with depression or normal controls. actigraphs in children who were victims of abuse, comparing their sleep Actigraphy has also been used to test treatment effects of melatonin ther- to that of children with major depression or dysthymia, and with that of apy in children with severe neurological disorders. 168 normal controls (evidence level 4C-b). In this study, abused children Taken as a whole, these clinical studies demonstrate the increasing had higher levels of nocturnal activity than both normal controls and experience in the use of actigraphy in a variety of populations, condi- depressed children, and the abused children had more difficulty falling tions and settings. Unfortunately, the majority of these studies do not and staying asleep. Hatonen et al. tested for differences in motor activi- report adequate detail on the technical aspects of the specific actigraph- ty rhythms with melatonin treatment versus placebo in 5 children aged ic devices used. However, it seems clear from these trials that the use of 12-19 with a neurodegenerative disease, neuronal ceroid lipofuscinosis actigraphy enables studies involving multiple days and nights of testing, 117 (NCL) (evidence level 5D-b). In these children, there were no differ- and allows populations that might otherwise not be studied, such as ences between melatonin and placebo in actigraphic motor activity patients with dementia or young children, to participate in research stud- based on period analysis by maximum entropy spectral, autocorrelation ies and clinical trials of sleep/wake activity and circadian rhythms or harmonic analyses. In a controlled trial of melatonin therapy in children with Rett syndrome (an X-linked genetic disorder with motor and 8.0 DISCUSSION cognitive impairment, and often severe sleep dysfunction), McArthur et al. found high variability in subject responsiveness to melatonin, but as The last published practice parameters for actigraphy only considered a group, sleep onset latency was significantly reduced with melatonin the use of actigraphy for the clinical assessment of sleep disorders.1 during the first 3 weeks of the trial (evidence level 4C-b).169 Their conclusion in 1995 was that actigraphy should not be used for the Mennella et al. tested the immediate, short-term effects of ethanol in clinical diagnosis of any sleep disorder, but that it might be a useful breast milk on actigraphy estimation of sleep in infants from 15 mother- adjunct to a good history and examination, particularly if multiple days infant pairs (evidence level 4C-b).170 They found lower total sleep time, of information were needed, if objective data on the pattern of sleep was lower active sleep and shorter sleep bouts when infants received breast needed or in order to clarify the effects of compliance with treatment.2 milk with ethanol compared to breast milk without ethanol. Since that time advances have been made in actigraphs and in the Sadeh and colleagues, in two separate studies, used actigraphy to algorithms that process their data both within the apparatus and on com- measure sleep in relation to cognitive functioning in preterm infants and puters following downloading of the data. Additionally, over 210 articles school-age children.171,172 In the first study, early mature patterns of and case studies have been published which have further examined the sleep were related to later cognitive maturity (evidence level 4C-b).171 In validity of actigraphy. As summarized in a recent review article by the second study, children with fragmented sleep had lower performance Sadeh and Acebo,173 the number of yearly publications on sleep and on measures of neurobehavioral functioning, particularly among the actigraphy has risen steadily in the last ten years. In the research setting, more complex tasks and behavior problems were more prevalent among actigraphs have been used for studying sleep disorders and circadian poor sleepers (evidence level 4C-b).172 rhythms. Actigraphic variables have also been used as outcome measures in clinical trials, often as a replacement for the more traditional, but 7.8 Summary more expensive and cumbersome PSG. One consistent finding of the current studies was that, when compared Actigraphy is increasingly being used in clinical research involving to PSG, actigraphy was found to be moderately valid and reliable for dif- individuals of various ages, who are of normal health or with a variety ferentiating sleep from wake in normal, healthy adult populations but of health conditions, and in a number of different settings. In the major- less reliable for identifying sleep as sleep became more disturbed. Taken ity of these studies, actigraphy was used to measure sleep and activity together, these studies provide evidence that important applications of rhythms that might not otherwise be available using traditional (e.g., the actigraph may be in the assessment and in the measurement of the PSG) techniques. In a growing number of sleep intervention trials, actig- sleep variability found in patients with insomnia, in assisting in the diag- raphy performed for multiple days and nights of testing was reported to nosis of circadian rhythm disorders, in characterizing and monitoring show evidence of beneficial treatment effects. Actigraphy has also been circadian rhythm disturbances that often accompany psychiatric disor- used in studies involving otherwise healthy adults to demonstrate sedat- ders, in studying sleep/wake patterns in populations where PSG would ing effects of various medications and to show differences in sleep dur- be difficult if not impossible, and in the assessment of treatment effects ing periods of sleep deprivation, for example, among military aircraft and follow-up studies. personnel on long flights. In addition, several large studies have used It is important to remember that actigraphy is not polysomnography. actigraphy in community-based samples to demonstrate differences Although actigraphy may not be 100% accurate when compared to PSG, between individuals based on age, gender, ethnicity, depressed mood, one still can get reliable information in situations where PSG is not prac- and other characteristics. tical. Actigraphy makes home recordings more accessible, permitting the Another growing area of research is cancer-related fatigue, where evaluation of patients in their natural sleeping environment and mini- studies involving multiple days and nights of actigraphy have demon- mizing laboratory effects that may alter a patient’s typical sleep patterns. strated that cancer patients with more robust circadian rhythms of activ- Actigraphy may provide an opportunity for subjects to adhere more ity report less fatigue, better quality of life and have longer survival. closely to their scheduled bedtime and wakeup time than a PSG record- Likewise, in a series of studies involving adults undergoing coronary ing, thus providing a more accurate estimate of typical sleep duration artery bypass surgery, the strength of circadian activity rhythms as mea- than does PSG. sured by actigraphy in the post-operative period was related to recovery

SLEEP, Vol. 26, No. 3, 2003 353 Actigraphy Review Paper—Ancoli-Israel et al LIMITATIONS answered. The data suggest that the wrist in general is more accurate for sleep estimation than other placements. Although traditionally actigra- In general, when actigraphy was compared to PSG, it was found to be phy has been recorded from the non-dominant wrist, newer data suggest both somewhat valid and apparently reliable in normal, healthy adult that movement from the dominant wrist may reflect sleep and wake populations. Overall, actigraphy is best at estimating total sleep time. more accurately than movements recorded from the non-dominant wrist. However, as sleep became more fragmented, the actigraph became less The problem of overestimating sleep must also be addressed. A poten- accurate in the detection of sleep and wake. Newer studies agree with tial approach to this problem might be the development of separate algo- 11 older studies (e.g., Webster et al., ) in suggesting that actigraphy may rithms for scoring sleep from daytime vs. nighttime activity records. It overestimate sleep and thus underestimate wake, particularly during the would be desirable for future research to systematically evaluate these or day when an individual is more likely to sit quietly while awake. In an other ways to overcome the actigraph’s tendency to overestimate sleep. effort to reduce this error, early investigators developed secondary algo- It may be that a different scoring algorithm is needed for periods when rithms that rescore sleep epochs as wake if adjacent to many wake the individual is expected to be awake (out-of-bed periods) than the one 11 epochs. currently used for periods during which the individual is expected to be While some data suggest that actigraphy consistently yields estimates asleep (in-bed periods). Studies of actigraphy compared to EEG outside of total sleep time and the number of awakenings that are higher than of the traditional sleep period, in patients that are known to fall asleep estimates on sleep logs, these results are more difficult to interpret, par- during the day, need to be done to more reliably determine the effec- ticularly since sleep logs themselves do not correlate highly with data tiveness of actigraphy during waking hours. from PSG. It is not unusual for clinicians to report that their patients are seen filling out a week’s worth of sleep logs while in the waiting room, SUMMARY waiting to be evaluated. This brings up two questions. First, what is the meaning of a reliability study when the comparison is made with some- In summary, although actigraphy is not as accurate as PSG for deter- thing other than a gold standard, such as when actigraphy is compared to mining some sleep measurements, studies are in general agreement that sleep logs? For example, although the actigraphic estimates of total actigraphy, with its ability to record continuously for long time periods, sleep time may have been higher than that reported in sleep logs, how do is more reliable than sleep logs which rely on the patients’ recall of how we know that the actigraph estimations were not actually more accurate many times they woke up or how long they slept during the night and is than the sleep log estimation? more reliable than observations which only capture short time periods. The second question is which is more important, the subjective report Actigraphy can provide information obtainable in no other practical way. of the sleep log or the objective estimation of actigraphy? Many clini- It can also have a role in the medical care of patients with sleep disor- cal trials are now using subjective reports as their final outcome mea- ders. However, it should not be held to the same expectations as sures, particularly in studies of insomnia, as they believe that if the polysomnography. Actigraphy is one-dimensional, whereas polysomno- patients feel they are sleeping better, it may not matter what the objec- graphy comprises at least 3 distinct types of data (EEG, EOG, EMG), tive data show. If that is the case, then neither actigraphy nor PSG are which jointly determine whether a person is asleep or awake. It is there- necessary. If, however, a more objective estimate is desired, then actig- fore doubtful whether actigraphic data will ever be informationally raphy is a less expensive approach than PSG and has the added benefit equivalent to the PSG, although progress on hardware and data process- of being able to record for multiple days and nights. When PSG is not ing software is continuously being made. feasible, the best approach may be to use a combination of actigraphy Although the 1995 practice parameters paper determined that actigra- with sleep logs. When there is agreement between the two methods, con- phy was not appropriate for the diagnosis of sleep disorders, more recent fidence is increased in the results of both. When there is disagreement, studies suggest that for some disorders, actigraphy may be more practi- it may reveal problems with one or the other. cal than PSG. While actigraphy is still not appropriate for the diagnosis Other problems with actigraphy relate to the determination of sleep of sleep disordered breathing or of periodic limb movements in sleep, it onset latency and variables whose calculations depend on it, for exam- is highly appropriate for examining the sleep variability (i.e., night-to- ple, sleep efficiency and wake after sleep onset. First, it is impossible to night variability) in patients with insomnia. Actigraphy is also appropri- determine sleep onset latency accurately without either an accurate ate for the assessment of and stability of treatment effects of anything marker of bedtime, such as a very accurate sleep log or an event marker from hypnotic drugs to light treatment to CPAP, particularly if assess- pushed at lights out. Activity monitors coupled with light sensors may be ments are done before and after the start of treatment. A recent indepen- useful for objectively determining the time of lights out, although a bed- dent review of the actigraphy literature by Sadeh and Acebo reached partner may continue to use lights after the person wearing the actigraph many of these same conclusions.173 has gone to sleep. Second, studies to date have often reported poor Some of the research studies failed to find relationships between sleep agreement between sleep onset latency estimated by actigraphy and that measures and health-related symptoms. The interpretation of these data determined by EEG. The problem may lie in the method of scoring, how- is also not clear-cut. Is it that the actigraph is not reliable enough to the ever, and not in the intrinsic properties of actigraphy. Cole et al.28 found access the relationship between sleep changes and quality of life mea- that the actigraph was more accurate (i.e., had a higher correlation) for sures, or, is it that, in fact, there is no relationship between sleep in that identifying sleep onset latency than any other sleep variable, when an population and quality of life measures? Other studies of sleep disor- appropriate scoring algorithm was used (see section 4.3. Comparisons to dered breathing, where actigraphy was not used and was not an outcome PSG). If this finding could be replicated, it might be possible to sub- measure also failed to find any relationship with quality of life. Is it then stantially improve actigraphic estimates of sleep latency, sleep efficien- the actigraph that is not reliable or that the associations just do not exist? cy and wake time after sleep onset. The one area where actigraphy can be used for clinical diagnosis is in the evaluation of circadian rhythm disorders. Actigraphy has been shown FUTURE RESEARCH to be very good for identifying rhythms. Results of actigraphic recordings correlate well with measurements of melatonin and of core body Standardization of acceptable norms needs to be established before temperature rhythms. Activity records also show sleep disturbance when actigraphy can be more generally used with full confidence in the realm sleep is attempted at an unfavorable phase of the circadian cycle. Actig- of sleep/wake studies. More development and research of both the raphy therefore would be particularly good for aiding in the diagnosis of devices that record the data and the algorithms that process the data is delayed or advanced sleep phase syndrome, non-24-hour-sleep syn- needed. In addition, disclosure of types of algorithms used should be drome and in the evaluation of sleep disturbances in shift workers. It required in all manuscripts. must be remembered, however, that overt rest-activity rhythms are sus- The question of the best placement of the actigraph must also be ceptible to various masking effects, so they may not always show the

SLEEP, Vol. 26, No. 3, 2003 354 Actigraphy Review Paper—Ancoli-Israel et al underlying rhythm of the endogenous circadian pacemaker. atry 1999; 9(4):203-210. In conclusion, the latest set of research articles suggest that in the clin- 31. Monk T, Buysse D, Rose LR. Wrist actigraphic measures of sleep in space. Sleep 1999; 22(7):948-954. ical setting, actigraphy is reliable for evaluating sleep patterns in patients 32. Dijk DJ, Wyatt JK, Wyatt JK et al. Sleep, performance, circadian rhythms, and light-dark with insomnia, for studying the effect of treatments designed to improve cycles during two space shuttle flights. American journal of physiology Regulatory, inte- sleep, in the diagnosis of circadian rhythm disorders (including shift grative and comparative physiology 2001; 281(5):R1647-R1664. 33. Sadeh A. Evaluating night wakings in sleep-disturbed infants: a methodological study of work), and in evaluating sleep in individuals who are less likely to tol- parental reports and actigraphy. Sleep 1996; 19(10):757-762. erate PSG, such as infants and demented elderly. While actigraphy has 34. Tikotzky L, Sadeh A. Sleep patterns and sleep disruptions in kindergarten children. Jour- been used in research studies for many years, up to now, methodological nal of Clinical and Child Psychology 2001; 30(4):581-591. issues had not been systematically addressed in clinical research and 35. Sadeh A. Assessment of intervention for infant night waking: Parental reports and activity-based home monitoring. J Consult Clin Psychol 1994; 62(1):63-68. practice. Those issues have now been addressed and actigraphy may 36. Acebo C, Sadeh A, Seifer R et al. Estimating sleep patterns with activity monitoring in now be reaching the maturity needed for application in the clinical arena. children and adolescents: How many nights are necessary for reliable measures? Sleep 1999; 22(1):95103. 37. Roehrs T, Turner L, Roth T. Effects of sleep loss on waking actigraphy. Sleep 2000; 9.0 REFERENCES 23(6):793-797. 38. Kazenwadel J, Pollmacher T, Trenkwalder C et al. New actigraphic assessment method 1. Sadeh A, Hauri PJ, Kripke DF, Lavie P. The Role of Actigraphy in the Evaluation of for periodic leg movements (PLM). Sleep 1995; 18(8):689-697. Sleep Disorders. Sleep 1995; 18(4):288-302. 39. Sforza E, Zamagni M, Petiav C, Krieger J. Actigraphy and leg movements during sleep: 2. Standards of Practice Committee A. An American Sleep Disorders Association Report: a validation study. J Clin Neurophysiol 1999; 16(2):154-160. Practice parameters for the use of actigraphy in the clinical assessment of sleep disor- 40. Middelkoop HA, Van Dam E, Smilde-van den Doel DA, Van Dijk G. 45-hour continu- ders. Sleep 1995; 18:285-287. ous quintuple-site actimetry: relations between trunk and limb movements and effects of 3. Broughton R, Fleming J, Fleetham J. Home assessment of sleep disorders by portable circadian sleep-wake rythmicity. Psychophysiol 2002; 34(2):199-203. monitoring. J Clin NeuroPhysiol 1996; 13(4):272-284. 41. Middelkoop HA, Van Dam E, Smilde-van den Doel DA. 45-hour multiple-site actime- 4. Teicher MH. Actigraphy and motion analysis: New tools for psychiatry. Harvard Review try in 20 healthy subjects: relations between body and limb movements and the effects of Psychiatry 1995; 3(1):18-35. of circadian sleep-wakefulness. Sleep-Wake Research in the Netherlands 1995;681-684. 5. Ancoli-Israel S. Actigraphy. In: Kryger MH, Roth T, Dement WC, editors. Principles and 42. Violani C, Testa P, Casagrande M. Actigraphic motor asymmetries during sleep. Sleep Practice of Sleep Medicine. Philadelphia: W.B. Saunders, 2000: 1295-1301. 1998; 21(5):472-476. 6. Kupfer DJ, Weiss BL, Foster FG, Detre TP, Delgado J, McPartland RJ. Psychomotor 43. Youngstedt SD, Kripke DF, Elliott JA, Klauber MR. Circadian abnormalities in older activity in affective states. Arch Gen Psychiatry 1974; 30:765-768. adults. J Pineal Res 2001; 31:264-272. 7. McPartland RJ, Kupfer DJ, Foster FG. The movement-activated recording monitor: A 44. Cole RJ, Smith JS, Alcala YC, Elliott JA, Kripke DF. Bright-light mask treatment of third generation motor-activity monitoring system. Behavior Research Methods & delayed sleep phase syndrome. J Biol Rhythms 2002; 17(1):89-101. Instrumentation 1976; 8:357-360. 45. Minors D, Akerstedt T, Atkinson G et al. The difference between activity when in bed 8. Colburn TR, Smith BM, Cuarini JJ, Simmons NM. An ambulatory activity monitor with and out of bed. I. Healthy subjects and selected patients. Chronobiol Intl 1996; 13:27- solid state memory. ISA Transactions 1976; 15:149-154. 34. 9. Kripke DF, Mullaney DJ, Messin S, Wyborney VG. Wrist actigraphic measures of sleep 46. Jean-Louis G, Kripke D, Mason WJ, Ancoli-Israel S. Relationships among illumination, and rhythms. Electroencephalogr Clin Neurophysiol 1978; 44:674-76. activity and sleep patterns. In: Holick MF, Jung EG, editors. Biological Effects of Light. 10. Mullaney DJ, Kripke DF, Messin S. Wrist-actigraphic estimation of sleep time. Sleep Kluver Academic Publishers, 1999: 37-39. 1980; 3:83-92. 47. Rechtschaffen A, Kales A, (eds). A Manual of Standardized Terminology, Techniques 11. Webster JB, Kripke DF, Messin S, Mullaney DJ, Wyborney G. An activity-based sleep and Scoring System for Sleep Stages of Human Subjects. Los Angeles: Brain Informa- monitor system for ambulatory use. Sleep 1982; 5:389-99. tion Service, Brain Research Institute, UCLA, 1973. 12. Jean-Louis G, Kripke DF, Cole RJ, Assmus JD, Langer RD. Sleep detection with an 48. Karacan I, Williams RL, Littell RC, Sails PJ. Insomniacs: Unpredictable and idiosyn- acceleromether actigraph: comparisons with polysomnography. Physiol Behav 2001; cratic sleepers. In: Koella WP, Levin P, editors. Sleep: Physiology, Biochemistry, Phar- 72:21-28. macology, Clinical Implications. Basel: Karger, 1972: 120-132. 13. Redmond DP, Hegge FW. The design of human activity monitors. In: Scheving LE, Hal- 49. Hauri PJ, Wisbey J. Wrist actigraphy in insomnia. Sleep 1992; 15(4):293-301. berg F, Ehret CF, editors. Chronobiology and Chronobiological Engineering. Dordrecht: 50. Chambers MJ. Actigraphy and insomnia: A closer look Part I. Sleep 1994; 17(5):405- Martinus Nijhoff Publishers, 1987: 202-15. 408. 14. Van Someren EJW, Lazeron RH, Vonk BF. Wrist acceleration and consequences for acti- 51. Guilleminault C, Clerk A, Black J, Labanowski M, Pelayo R, Claman D. Nondrug treat- graphic rest-activity registration in young and elderly subjects. Sleep-Wake Research in ment trials in psychophysiologic insomnia. Ann Intern Med 1995; 155:838-844. the Netherlands 1995;6123-6125. 52. Wilson KG, Watson ST, Currie SR. Daily diary and ambulatory activity monitoring of 15. Gorny SW, Spiro JR. Comparing different methodologies used in wrist actigraphy. Sleep sleep in patients with insomnia associated with chronic musculoskeletal pain. Pain 1998; Review 2001; Summer:40-42. 75(1):75-84. 16. Jean-Louis G, von Gizycki H, Zizi F, Spielman A, Hauri P, Taub H. The actigraph data 53. Jean-Louis G, Kripke D, Ancoli-Israel S. Sleep and quality of well-being. Sleep 2000; analysis software: I. A novel approach to scoring and interpreting sleep-wake activity. 23(8):1115-1121. Percept Mot Skills 1997; 85(1):207-216. 54. Dagan Y, Zinger Y, Lavie P. Acitgraphic sleep montoring in post-traumatic stress disor- 17. Sadeh A, Sharkey KM, Carskadon MA. Activity-based sleep-wake identification: An der (PTSD) patients. J Psychosom Res 1997; 42(6):577-581. empirical test of methodological issues. Sleep 1994; 17:201-207. 55. Wicklow A, Espie CA. Intrusive thoughts and their relationship to actigraphic measure- 18. Pollak CP, Stokes PE, Wagner DR. Direct comparision of two widely used activity ment of sleep: Towards a cognitive model of insomnia. Behav Res Ther 2000; 38:679- recorders. Sleep 1998; 21(2):207-212. 693. 19. Kushida CA, Chang A, Gadkary C, Guilleminault C, Carrillo O, Dement WC. Compar- 56. Kerkhof G, Van Vianen B. Circadian phase estimation of chronic insomniacs relates to ison of actigraphic, polysomnographic, and subjective asessment of sleep parameters in their sleep characteristics. Archives of Physiology and Biochemistry 1999; 107(5):383- sleep-disordered patients. Sleep Med 2001;398-396. 392. 20. Tryon WW. Activity and sleep. Act Measure Psychol Med 1991;149-195. 57. Dagan Y, Yovel I, Hallis D, Eisenstein M, Raichik I. Evaluating the role of melatonin in 21. Pollak C, Tryon WW, Nagaraja H, Dzwonczyk R. How accurately does wrist actigraphy the long-term treatment of delayed sleep phase syndrome (DSPS). Chronobiol Int 1998; identify states of sleep and wakefulness? Sleep 2001; 24(8):957-965. 15(2):181-190. 22. Jean-Louis G, von Gizycki H, Zizi F et al. Determination of sleep and wakefulness with 58. Okawa M, Urata J, Uchiyama M et al. Melatonin treatment for circadian rhythm sleep the actigraph data analysis software (ADAS). Sleep 1996; 19(9):739-743. disorders. Psychiatry and Clinical Neurosciences 1998; 52(2):259-260. 23. Blood ML, Sack RL, Percy DC, Pen JC. A comparison of sleep detection by wrist actig- 59. Quinto C, Gellido C, Chokroverty S, Masdeu J. Posttraumatic delayed sleep phase syn- raphy, behavioral response, and polysomnography. Sleep 1997; 20(6):388-395. drome. Neurology 2000; 54(1):250-252. 24. Jean-Louis G. Sleep estimation from wrist movement quantified by different actigraph- 60. Nagtegaal J, Kerkhof GA, Smits MG, Swart AC, Van Der Meer YG. Traumatic brain ic modalities. J Neurosci Methods 2001; 105(2):185-191. injury-associated delayed sleep phase syndrome. Functional Neurology 1997; 25. Reid K, Dawson D. Correlation between wrist activity monitor and electrophysiological 12(6):345-348. measures of sleep in a simulated shiftwork environment for younger and older subjects. 61. Nagtegaal J, Kerkhof GA, Smits MG, Swart AC, Van Der Meer YG. Delayed sleep phase Sleep 1999; 52(2):157-159. syndrome: a placebo-controlled cross-over study on the effects of melatonin adminis- 26. Ancoli-Israel S, Clopton P, Klauber MR, Fell R, Mason WJ. Use of wrist activity for tered five hours before the individual dim light melatonin onset. J Sleep Res 1998; monitoring sleep/wake in demented nursing home patients. Sleep 1997; 20(1):24-27. 7(2):135-143. 27. Shinkoda H, Matsumoto K, Hamasaki J, Seo YJ, Park YM, Park KP. Evaluation of 62. Franck LS, Johnson LM, Lee K et al. Sleep disturbances in children with human immun- human activites and sleep-wake identification using wrist actigraphy. Psychiatry 1998; odeficiency virus infection. Pediatrics 1999; 104(5). 52(2):157-159. 63. Hering E, Epstein R, Elroy S, Iancu DR, Zelnik N. Sleep patterns in autistic children. 28. Cole RJ, Kripke DF, Gruen W, Mullaney DJ, Gillin JC. Automatic sleep/wake identifi- Journal of Autism and Developmental Disorders 1999; 29(2):143-147. cation from wrist activity. Sleep 1992; 15(5):461-469. 64. Sadeh A, Dark I, Vohr BR. Newborns’sleep-wake patterns: The role of maternal, deliv- 29. Lockley SW, Skene DJ, Arendt J. Comparison between subjective and actigraphic mea- ery and infant factors. Early Hum Dev 1996; 44:113-126. surement of sleep and sleep rhythms. J Sleep Res 1999; 8(3):175-184. 65. Etzioni A, Luboshitzky R, Tiosano D, Ben-Harush M, Goldsher D, Lavie P. Melatonin 30. Krahn L, Lin SC, Wisbey J, Rummans T, O’Connor MK. Assessing sleep in psychiatric replacement corrects sleep disturbances in a child with pineal tumor. Neurology 1996; inpatients: nurse and patient reports versus wrist actigraphy. Annals of Clinical Psychi-

SLEEP, Vol. 26, No. 3, 2003 355 Actigraphy Review Paper—Ancoli-Israel et al 46(1):261-263. of morning bright light therapy for rest-activity rhythm disorders in patients with vascu- 66. Sadeh A, Raviv A, Gruber R. Sleep patterns and sleep disruptions in school-age children. lar dementia and dementia of Alzheimer’s type. Chronobiol Int 1998; 15(6):647-654. Developmental Psychology 2000; 36:291-301. 100. van Someren E, Kessler A, Mirmiran M, Swaab DF. Indirect bright light improves cir- 67. Aronen ET, Paavonen EJ, Fjallberg M, Soininen M, Torronen J. Sleep and psychiatric cadian rest-activity rhythm disturbances in demented patients. Biol Psychiat 1997; symptoms in school-age children. J Am Acad Child Adolesc Psychiatry 2000; 39(4):502- 41:955-963. 508. 101. van Someren EJ, Lijzenga C, Mirmiran M, Swaab DF. Long-term fitness training 68. Kramer CJ, Kerkhof G, Hofman WF. Age differences in sleep-wake behavior under nat- improves the circadian rest-activity rhythm in healthy elderly males. J Biol Rhythms ural conditions. Personality and Individual Differences 1999; 27:853-860. 1997; 12(2):146-156. 69. Middelkoop HA, Knuistingh NA, van Hilten JJ, Ruwhof CW, Kamphuisen HA. Wrist 102. Yoon IY, Jeong DU, Kwon KB, Kang SB, Song BG. Bright light exposure at night and actigraphic assessment of sleep in 116 community based subjects suspected of obstruc- light attenuation in the morning improve adaptation of night shift workers. Sleep 2002; tive sleep apnoea syndrome. Thorax 1995; 50(3):284-289. 25(3):351-356. 70. Drinnan MJ, Murray A, White JES, Smithson AJ, Gibson GJ, Griffiths CJ. Evaluation of 103. Luna TD, French J, Mitcha JL. A study of USAF air traffic controller shiftwork: sleep, activity-based techniques to identify transient arousal in respiratory sleep disorders. J fatigue, activity, and mood analyses. Aviat Space Environ Med 1997; 68(1):18-23. Sleep Res 1996; 5:173-180. 104. Thessing VC, Anch AM, Muehlbach MJ, Schweitzer PK, Walsh JK. Two- and 4-hour 71. Elbaz M, Roue GM, Lofaso F, Quera-Salva M. Utility of actigraphy in the diagnosis of bright-light exposures differnetially effect sleepiness and performance the subsequent obstructive sleep apnea. Sleep 2002; 25(5):527-531. night. Sleep 1994; 17(2):140-143. 72. Trenkwalder C, Stiasny K, Pollmacher T et al. L-Dopa therapy of uremic and idiopath- 105. Jockovich M, Cosentino D, Cosentino L, Wears RL, Seaberg DC. Εffect of exogenous ic restless legs syndrome: A double-blind, crossover trial. Sleep 1995; 18(8):681-688. melatonin on mood and sleep efficiency in emergency medicine residents working night 73. Collado-Seidel V, Kazenwadel J, Wetter TC et al. A controlled study of additional sr-L- shifts. Acad Emerg Med 2000; 7(8):955-958. dopa in L-dopa-responsive restless legs syndrome with late-night symptoms. Neurology 106. Park YM, Matsumoto K, Seo YJ, Cho YR, Noh TJ. Sleep-wake behavior of shift work- 1999; 52(2):285-290. ers using wrist actigraph. Psychiatry and Clinical Neurosciences 2000; 54(3):359-360. 74. Plazzi G, Schultz Y, Cortelli P et al. Motor overactivity and loss of motor circadian 107. Glod CA, Teicher MH, Polcari A, McGreenery CE, Ito Y. Circadian rest-activity distur- rhythm in fatal familial insomnia: an actigraphic study. Sleep 1997; 20(9):739-742. bances in children with seasonal affective disorder. J Am Acad Child Adoles Psychiatry 75. Webb WB, Rivera DL. A case of extremely long sleep and waking episodes. Sleep 1994; 1997; 36(2):199-195. 17(7):646-649. 108. Sakurai N, Sasaki M. An activity monitor study on the sleep-wake rhythm of healthy 76. Shibui K, Uchiyama M, Iwama H, Ozaki S, Takahashi K, Okawa M. Periodic fatigue aged people residing in their homes. Psychiatry and Clinical Neurosciences 1998; symptoms due to desynchronization in a patient with non-24-h sleep-wake syndrome. 52(2):253-255. Psychiatry 1998; 52(5):477-481. 109. Martin J, Marler MR, Shochat T, Ancoli-Israel S. Circadian rhythms of agitation in insti- 77. Kunz D, Bes F. Melatonin effects in a patient with severe REM sleep behavior disorder: tutionalized patients with Alzheimer’s Disease. Chronobiol Intl 2000; 17(3):405-418. case report and theoretical considerations. Neuropsychobiology 1997; 34(4):211-214. 110. Ancoli-Israel S, Martin JL, Kripke DF, Marler M, Klauber MR. Effect of light treatment 78. Frisbie JH, Sharma GVRK. Circadian rhythm of pulmonary embolism in patients with on sleep and circadian rhythms in demented nursing home patients. J Am Geriatr Soc acute spinal cord injury. Amer J Cardiol 1992; 70:827. 2002; 50(2):282-289. 79. Haug H-J, Wirz-Justice A, Rossler W. Actigraphy to measure day structure as a thera- 111. van Someren E, Swaab DF, Colenda CC, Cohen W, McCall WV, Rosenquist PB. Bright peutic variable in the treatment of schizophrenic patients. Acta Psychiat Scand 2000; light therapy: improved sensitivity to its effects on rest-activity rhythms in Alzheimer’s 102:91-95. patients by application of nonparametric disorder. Chronobiol Int 1999; 16(4):505-518. 80. Werth E, Savaskan E, Knoblauch V et al. Decline in long-term circadian rest-activity 112. Ancoli-Israel S, Klauber MR, Jones DW et al. Variations in circadian rhythms of activ- cycle organization in a patient with dementia. J Geriatr Psychiatry Neurol 2002; 15:55- ity, sleep and light exposure related to dementia in nursing home patients. Sleep 1997; 59. 20(1):18-23. 81. Wirz-Justice A, Haug H-J, Cajochen C. Disturbed circadian rest-activity cycles in 113. Gruber R, Sadeh A, Raviv A. Instability of sleep patterns in children with attention- schizophrenic patients: An effect of drugs? Schizophrenia Bul 2001; in press. deficit/heperactivity disorder. J Am Acad Child Adolesc Psychiat 2000; 39(4):495-501. 82. Wirz-Justice A, Quinto C, Cajochen C, Werth E, Hock C. A rapid-cycling bipolar patient 114. Scherder EJA, van Someren EJ, Swaab DF. Transcutaneous electrical nerve stimulation treated with long nights, bedrest, and light. Biol Psychiatry 1999; 45(8):1075-1077. (TENS) improves the rest-activity rhythm in midstage Alzheimer’s disease. Behav Brain 83. Wirz-Justice A, Cajochen C, Nussbaum P. A schizophrenic with an arrhythmic circadian Res 1999; 101(1):105-107. rest-activity cycle. Psychiatry Res 1997; 73:83-90. 115. van Someren EJ, Hagebeuk EEO, Lijzenga C et al. Circadian rest activity rhythm dis- 84. Wirz-Justice A, Werth E, Savaskan E, Knoblauch V, Gasio PF, Muller-Spahn F. turbances in Alzheimer’s disease. Biol Psychiatry 1996; 40(4):259-270. Haloperidol disrupts, clozapine reinstates the circadian rest-activity cycle in a patient 116. Huang Y-L, Liu R-Y, Want Q-S, van Someren EJ, Xu H, Zhou J-N. Age-associated dif- witih early-onset Alzheimer’s disease. Alzheimer Disease and Associated Disorders ference in circadian sleep-wake and rest-activity rhythms. Physiol Behav 2002; 76:597- 2002; 14(4):212-215. 603. 85. Siegmund R, Tittel M, Schiefenhovel W. Activity monitoring of the inhabitants in 117. Hatonen T, Kirveskari E, Heiskala H et al. Melatonin ineffective in neuronal ceroid lipo- Tauwema, a traditional Melanesian village: rest/activity behaviour of Trobriand fuscinosis patients with fragmented or normal motor activity rythms recorded by wrist islanders (Papua New Guinea). Biological Rhythm Research 1998;2949-2959. actigraphy. Molecular Genetics and Metabolism 1999; 66(4):401-406. 86. Scher A, Tirosh E, Lavie P. The relationship between sleep and temperament revisited: 118. Carskadon MA, Acebo C, Richardson GS, Tate BA, Seifer R. An approach to studying evidence for 12-month-olds: a research note. J Child Psychol Psychiatry 1998; circadian rhythms of adolescent humans. J Biol Rhythms 1997; 12(3):278-289. 39(5):785-788. 119. Carskadon MA, Wolfson A, Acebo C, Tzischinsky O, Seifer R. Adolescent sleep pat- 87. Blagrove M, Owens DS, MacDonald I, Sytnik N, Tucker P, Folkard S. Time of day terns, circadian timing, and sleepiness at a transition to early school days. Sleep 1998; effects in, and the relationship between, sleep quality and movement. J Sleep Res 1998; 21(8):871-881. 7(4):233-239. 120. Heikkila E, Hatonen TH, Telakivi T et al. Circadian rhythm studies in neuronal ceroid- 88. Evans BD, Rogers AE. 24-hours sleep/wake patterns in healthy elderly persons. Appl lipofuscinosis (NCL). Amer J Med Genet 1995; 57:229. Nurs Res 1994; 7(2):75-83. 121. Guilleminault C, Leger D, Pelayo R, Gould S, Hayes B, Miles L. Development of cir- 89. Middleton B, Arendt J, Stone BM. Complex Effects of Melatonin on Human Circadian cadian rhythmicity of temperature in full-term normal infants. Neurophysiol Clin 1996; Rhythms in Constant Dim Light. J Bio Rhythms 1997; 12(5):467-477. 26(1):21-29. 90. Middleton B, Arendt J, Stone BM. Human circadian rhythms in constant dim light (8 122. Jean-Louis G, Mendlowicz MV, von Gizycki H, Zizi F, Nunes J. Assessment of physi- lux) with knowledge of clock time. J Sleep Res 1996; 5(2):69-76. cal activity and sleep by actigraphy: examination of gender differences. Journal of Wom- 91. Satlin A, Teicher MH, Lieberman HR, Baldessarini RJ, Volicer L, Rheaume Y. Circadi- ens Health & Gender-Based Medicine 1999; 8(8):1113-1117. an locomotor activity rhythms in Alzheimer’s disease. Neuropsychopharmacol 1991; 123. Pollak CP, Sokes PE. Circadian rest-activity rhythms in demented and nondemented 5(2):115-126. older community residents and their caregivers. J Am Geriatr Soc 1997; 45:446-452. 92. Binkley S. Wrist motion rhythm phase shifts in travelers may differ from changes in time 124. Jean-Louis G, Kripke D, Ancoli-Israel S et al. Circadian sleep, illumination and activity zones. Physiology & Behavior 1994; 55:967. patterns in women: Influences of aging and time reference. Physiology & Behavior 93. Dawson D, Encel N, Lushington K. Improving adaptation to simulated night shift: 2000; 68:347-352. Timed exposure to bright light versus daytime melatonin administration. Sleep 1995; 125. Mishima K, Okawa M, Satoh K, Shimizu T, Hozumi S, Hishikawa Y. Different mani- 18(1):11-21. festations of circadian rhythms in senile dementia of Alzheimer’s type and multi-infarct 94. Quera-Salva MA, Defrance R, Claustrat B, De Lattre J, Guilleminault C. Rapid shift in dementia. Neurobiol Aging 1997; 18(1):105-109. sleep time and acrophase of melatonin secretion in short shift work schedule. Sleep 126. Friedman L, Kraemer HC, Zarcone V et al. Disruptive behavior and actigraphic mea- 1996; 19(7):539-543. sures in home-dwelling patients with Alzheimer’s Disease: Preliminary Report. J Geri- 95. Quera-Salva MA, Guilleminault C, Claustrat B et al. Rapid shift in peak melatonin atr Psychiatry Neurol 1997; 10:58-62. secretion associated with improved performance in short shift work schedule. Sleep 127. Shapiro D, Goldstein I. Wrist actigraph measures of physical activity level and ambula- 1997; 20(12):1145-1150. tory blood pressure in healthy elderly persons. Psychophysiol 1998; 35(3):305-312. 96. Minors D, Waterhouse J, Folkard S, Atkinson G. The difference between activity when 128. Mansoor GA, White WB, McCabe EJ, Giacco S. The relationship of electronically mon- in bed and out of bed. III. Nurses on night work. Chronobiol Intl 1996; 13(4):273-282. itored physical activity to blood pressure, heart rate, and the circadian blood pressure 97. Daurat A, Benoit O, Buguet A. Effects of zopiclone on the rest/activity rhythm after a profile. Am J Hypertension 2000; 13(3):262-267. westward flight across five time zones. Psychopharmacology (Berlin) 2000; 149(3):241- 129. Leary AC, Donnan PT, MacDonald TM, Murphy MB. Physical activity level is an inde- 245. pendent predictor of the diurnal variation in blood pressure. J Hyperten 2000; 18(4):405- 98. Kubota T, Uchiyama M, Hirokawa G, Ozaki S, Hayasi M, Okawa M. Effects of evening 410. light on body temperature. Psychiatry and Clinical Neurosciences 1998; 52(2):248-249. 130. White WB, Mansoor GA, Pickering TG et al. Differential effects of morning and 99. Mishima K, Hishikawa Y, Okawa M. Randomized, dim light controlled, crossover test evening dosing of nisoldipine ER on circadian blood pressure and heart rate. Am J

SLEEP, Vol. 26, No. 3, 2003 356 Actigraphy Review Paper—Ancoli-Israel et al Hypertension 1999; 12:806-814. nursing home residents: association with sleep and bed mobility. J Am Geriatr Soc 1995; 131. Eissa MA, Yetman RJ, Poffenbarger T, Portman RJ. Comparison of arbitrary definitions 43:788-792. of circadian time periods with those determined by wrist actigraphy in analysis of ABPM 164. Alessi CA, Yoon EJ, Schnelle JF, Al-Samarrai NR, Cruise PA. A randomized trial of a data. Journal of Human Hypertension 1999; 13(11):759-763. combined physical activity and environmental intervention in nursing home residents: 132. Martin J, Jeste DV, Caligiuri M, Patterson T, Heaton R, Ancoli-Israel S. Actigraphic esti- Do sleep and agitation improve? J Am Geriatr Soc 1999; 47:784-791. mates of circadian rhythms and sleep/wake in older schizophrenia patients. Schizophr 165. Cruise PA, Schnelle JF, Alessi CA, Simmons SF, Ouslander JG. The nighttime environ- Res 2001; 47(1):77-86. ment and incontinence care practices in nursing homes. J Am Geriatr Soc 1998; 133. Poyurovsky M, Nave R, Epstein R et al. Actigraphic monitoring (actigraphy) of circa- 46(2):463-466. dian locomotor activity in schizophrenic patients with acute neuroleptic-induced 166. Ouslander JG, Buxton WG, Al-Samarrai NR, Cruise PA, Alessi CA, Schnelle JF. Night- akathisia. European Neuropsychopharmacology 2000; 10:171-176. time urinary incontinence and sleep disruption among nursing home residents. J Am 134. Lemke MR, Broderick A, Zeitelberger M, Hartmann W. Motor activity and daily varia- Geriatr Soc 1998; 46(4):463-466. tion of symptom intensity in depressed patients. Neuropsychobiology 1997; 36(2):57- 167. Corkum P, Tannock R, Moldofsky H, Hogg-Johnson S, Humphries T. Actigraphy and 61. parental ratings of sleep in children with attention-deficit/hyperactivity disorder 135. Garfinkel D, Laudon M, Nof D, Zisapel N. Improvement of sleep quality in elderly peo- (ADHD). Sleep 2001; 24(3):303-312. ple by controlled-release melatonin. Lancet 1995; 346:541-544. 168. Glod CA, Teicher MH, Hartman CR, Harakal T. Increased nocturnal activity and 136. Friedman L, Benson K, Noda A et al. An actigraphic comparison of sleep restriction and impaired sleep maintenance in abused children. J Am Acad Child Adolesc Psychiat sleep hygiene treatments for insomnia in older adults. J Geriatr Psychiatry Neurol 2000; 1997; 36(9):1236-1243. 13(1):17-27. 169. McArthur AJ, Budden SS. Sleep dysfunction in Rett syndrome: a trial of exogenous 137. Maus TL, McLaren JW, Shepard JW, Jr., Brubaker RF. The effects of sleep on circulat- melatonin treatment. Dev Med Child Neurol 1998; 40(3):186-192. ing catecholamines and aqueous flow in human subjects. Exp Eye Res 1996; 62(4):351- 170. Mennella JA, Gerrish CJ. Effects of exposure to alcohol in mother’s milk on infant 358. sleep. Pediatrics 1998; 101 (5):Ε2. 138. Pollak CP, Perlick D, Linsner JP. Sleep and motor activity of community elderly who 171. Gertner S, Greenbaum CW, Sadeh A, Dolfin Z, Sirota L, Ben-Nun Y. Sleep-wake pat- frequently use bedtime medications. Biol Psychiatry 1994; 35:73-75. terns in preterm infants and 6 month’s home environment: implications for early cogni- 139. Duka T, Goerke D, Fichte K. Effects of ZK 93 426, a beta-carboline benzodiazepine tive development. Early Hum Dev 2002; 68(2):93-102. receptor antagonist on night sleep pattern in healthy male volunteers. Psychopharmacol- 172. Sadeh A, Gruber R, Raviv A. Sleep, neurobehavioral functioning, and behavior prob- ogy (Berlin) 1995; 117:178. lems in school-age children. Child Development 2002; 73(2):405-417. 140. French J, Bisson RU, Neville KJ, Mitcha JL, Storm WF. Crew fatigue during simulated, 173. Sadeh A, Acebo C. The role of actigraphy in sleep medicine. Sleep Medicine Reviews long duration B-1B bomber missions. Aviat Space Environ Med 1994; 65(5):A1-6. 2002; 6(2):113-124. 141. Hindmarch I, Rigney U, Stanley N, Quinlan P, Rycroft J, Lane J. A naturalistic investi- 174. Eissa MA, Poffenbarger T, Portman RJ. Comparison of the actigraph versus patients’ gation of the effects of day-long consumption of tea, coffee and water on alertness, sleep diary information in defining circadian time periods for analyzing ambulatory blood onset and sleep quality. Psychopharmacology (Berlin) 2000; 149(3):203-216. pressure monitoring data. Blood Pressure Monitoring 2001; 6(1):21-25. 142. Hindmarch I, Shamsi Z, Stanley N, Fairweather DB. A double-blind, placebo-controlled 175. Hauri PJ. Evaluation of a sleep switch device. Sleep 1999; 22(8):1110-1117. investigation of the effects of fexofenadine, loratadine and promethazine on cognitive 176. Jean-Louis G, von Gizycki H, Zizi F, Spielman A, Hauri P, Taub H. The actigraph data and psychomotor function. Brit J Clin Pharmacol 1999; 48(2):200-206. analysis software: II. A novel approach to scoring and interpreting sleep-wake activity. 143. Jean-Louis G, Kripke DF, Ancoli-Israel S, Klauber MR, Sepulveda RS. Sleep duration, Percept Mot Skills 1997; 85(1):219-226. illumination, and activity patterns in a population sample: Effects of gender and ethnic- 177. Jean-Louis G, Zizi F, von Gizycki H, Hauri P. Actigraphic assessment of sleep in insom- ity. Biol Psychiatry 2000; 47:921-927. nia: application of the Actigraph Data Analysis Software (ADAS). Physiol Behav 1999; 144. Mendlowicz MV, Jean-Louis G, von Gizycki H, Zizi F, Nunes J. Actigraphic predictors 65:569-563. of depressed mood in a cohort of non-psychiatric adults. Aust NZ J Psychiatry 1999; 178. Otsuka K, Cornelissen G, Halberg F. Broad scope of a newly developed actometer in 33(4):553-558. chronobiology, particularly chronocardiology. Chronobiologia 1994; 21:251. 145. Moorcroft WH, Kayser KH, Griggs AJ. Subjective and objective confirmation of the 179. van Someren EJ, Lazeron RH, Vonk BF, Mirmiran M, Swaab DF. Gravitational artefact ability to self-awaken at a self-predetermined time without using external means. Sleep in freqency spectra of movement acceleration: implications for actigraphy in young and 1997; 20(1):40-45. elderly subjects. J Neurosci Methods 1996; 65(1):55-62. 146. Pankhurst FP, Horne JA. The influence of bed partners on movement during sleep. Sleep 180. Verbeek I, Arends J, Declerck G, Beecher L. Wrist actigraphy in comparison with 1994; 17(4):308-315. polysomnography and subjective evaluation in insomnia. In: Coenen AML, editor. 147. Reyner A, Horne JA. Gender- and age-related differences in sleep determined by home- Sleep-Wake Research in The Netherlands. Leiden: Dutch Society for Sleep-Wake recorded sleep logs and actimetry from 400 adults. Sleep 1995; 18(2):127-134. Research, 1994: 163-170. 148. Berger AM, Farr L. The influence of daytime inactivity and nighttime restlessness on 181. Guilleminault C, Yuen KM, Gulevich MG, Karadeniz D, Leger D, Philip P. Hypersom- cancer-related fatigue. Oncol Nurs Forum 1999; 26:1663-1671. nia after head-neck trauma: a medicolegal dilemma. Neurology 2000; 54(3):653-659. 149. Miaskowski C, Lee KA. Pain, fatigue and sleep disturbances in oncology outpatients 182. Hayes MJ, Mitchell D. Spontaneous movements during sleep in children: temporal orga- receiving radiation therapy for bone metastasis: a pilot study. Journal of Pain Symptom nization and changes with age. Dev Psychobiol 1998; 32(1):13-21. Management 1999; 17(5):320-332. 183. Kunz D, Bes F. Exogenous melatonin in periodic limb movement disorder: an open clin- 150. Mormont MC, Waterhouse J, Bleuzen P et al. Marked 24-h rest/activity rhythms are ical trial and a hypothesis. Sleep 2001; 24(2):183-187. associated with better quality of life, better response and longer survival in patients with 184. Honma H, Kohsaka M, Suzuki I et al. Motor activity rhythm in dementia with delirium. metastatic colorectal cancer and good performance status. Clinical Cancer Research Psychiatry Clin Neurosci 1998; 52(2):196-198. 2000; 6:3038-3045. 185. Kario K, Schwartz JE, Pickering TG. Ambulatory physical activity as a determinant of 151. Berger AM. Patterns of fatigue and activity and rest during adjuvant breast cancer diurnal blood pressure variation. Hypertension 1999; 34:685-691. chemotherapy. Oncol Nurs Forum 1998; 25(1):51-62. 186. Lowden A, Akerstedt T. Retaining home-base sleep hours to prevent jet lag in connec- 152. Dursun SM, Patel JK, Burke JG, Reveley MA. Effects of typical antipsychotic drugs and tion with a westward flight across nine time zones. Chron Intl 2002; 15(4):365-376. risperidone on the quality of sleep in patients with schizophrenia: a pilot study. J Psy- 187. Luboshitzky R, Shen-Orr Z, Tzischinsky O, Maldonado M, Herer P, Lavie P. Acti- chiatr Neurosci 1999; 24(4):333-337. graphic sleep-wake patterns and urinary 6-sulfatoxymelatonin excretion in patients with 153. Lemke MR, Puhl P, Broderick A. Motor activity and perception of sleep in depressed Alzheimer’s disease. Chronobiol Int 2001; 18(3):513-524. patients. J Psychiatr Res 1999; 33(3):215-224. 188. Lushington K, Dawson D, Encel N, Lack L. Urinary 6-sulfatoxymelatonin cycle-to- 154. Cordoba J, Cabrera J, Lataif L, Penev P, Zee P, Blei AT. High prevalence of sleep dis- cycle variability. Chronobiol Int 1996; 13(6):441-421. turbance in cirrhosis. Hepatology 1998; 27(2):339-345. 189. Samel A, Vejvoda M, Wittiber K, Wenzel J. Joint NASA-ESA-DARA Study. Part three: 155. Redeker NS, Wykpisz E. Effects of age on activity patterns after coronary artery bypass circadian rhythms and activity-rest cycle under different CO2 concentrations. Aviat surgery. Heart Lung 1998; 28(1):5-14. Space Environ Med 1998; 69(5):501-505. 156. Redeker NS, Mason DJ, Wykpisz E, Glica B. Sleep patterns in women after coronary 190. van Someren EJ, Scherder EJA, Swaab DF. Transcutaneous electrical nerve stimulation artery bypass surgery. Appl Nurs Res 1996; 9(3):115-122. (TENS) improves circadian rhythm disturbances in Alzheimer disease. Alzheimer Dis 157. Redeker NS, Tamburri L, Howland CL. Prehospital correlates of sleep in patients hos- Assoc Disord 1998; 12(2):114-118. pitalized iwth cardiac disease. Res Nurs Health 1998; 21(1):27-37. 191. Baker A, Simpson A, Dawson D. Sleep disruption and mood changes associated with 158. Pollak CP, Stokes PE, Wagner DR. Nocturnal interactions between community elders menopause. J Psychosomatic Res 1997; 43(4):359-369. and caregivers, as measured by cross-correlation of their motor activity. J Geriatr Psy- 192. Dagan Y, zeevi-luria S, Sever Y et al. Sleep quality in children with attention deficit chiatry Neurol 1997; 10:168-173. hyperactivity disorder; An actigraphic study. Psychiatry and Clinical Neurosciences 159. Ancoli-Israel S, Jones DW, Hanger MA, Parker L, Klauber MR, Kripke DF. Sleep in the 1997; 51:383-386. nursing home. In: Kuna ST, Suratt PM, Remmers JE, editors. Sleep and Respiration in 193. Dinges DF, Pack F, Williams K et al. Cumulative sleepiness, mood disturbance, and psy- Aging Adults. New York: Elsevier Press, 1991: 77-84. chomotor vigilance performance decrements during a week of sleep restricted to 4-5 160. Pat-Horenczyk R, Klauber MR, Shochat T, Ancoli-Israel S. Hourly profiles of sleep and hours per night. Sleep 1997; 20(4):267-277. wakefulness in severely versus mild-moderately demented nursing home patients. Aging 194. Glod CA, Teicher MH. Relationship between early abuse, posttraumatic stress disorder, Clin Exp Res 1998; 10:308-315. and activity levels in prepubertal children. J Am Acad Child Adolesc Psychiatry 1996; 161. Shochat T, Martin J, Marler M, Ancoli-Israel S. Illumination levels in nursing home 35(10):1384-1393. patients: Effects on sleep and activity rhythms. J Sleep Res 2000; 9(4):373-380. 195. Haimov I, Lavie P, Laudon M, Herer P, Vigder C, Zisapel N. Melatonin replacement 162. Alessi CA, Schnelle JF, MacRae PG et al. Does physical activity improve sleep in therapy of elderly insomniacs. Sleep 1995; 18(7):598-603. impaired nursing home residents? J Am Geriatr Soc 1995; 43:1098-1102. 196. Humphreys JC, Lee KA, Neylan TC, Marmar CR. Sleep patterns of sheltered battered 163. Alessi CA, Schnelle JF, Traub S, Ouslander JG. Psychotropic medications in incontinent women. Image J Nurs Sch 1999; 31(2):139-143.

SLEEP, Vol. 26, No. 3, 2003 357 Actigraphy Review Paper—Ancoli-Israel et al 197. Ito T, Yamadera H, Ito R, Suzuki H, Asayama K, Endo S. Effects of vitamin B12 on BP - blood pressure bright light on cognitive and sleep–wake rhythm in Alzheimer-type dementia. Psychia- BRM - behavioral response monitoring try and Clinical Neurosciences 2001; 55:281-282. 198. Jean-Louis G, Mendlowicz MV, Gillin JC et al. Sleep estimation from wrist activity in BT - bedtime patients with major depression. Physiol Behav 2000; 70:49-53. CES-D - Center for Epidemiologic Studies-Depression scale 199. Jean-Louis G, von Gizycki H, Zizi F. Melatonin effects on sleep, mood, and cognition in CR - circadian rhythm elderly with mild cognitive impairment. J Pineal Res 2002; 25(3):177-183. 200. Kario K, Schwartz JE, Davidson KW, Pickering TG. Gender differences in associations CSQP – Child Sleep Questionnaire Parent of diurnal blood pressure variation, awake physical activity, and sleep quality with neg- D - day ative affect: the work site blood pressure study. Hypertension 2001; 38(5):997-1002. DB - double-blinded 201. Lee KA, Portillo CJ, Miramontes H. The influence of sleep and activity patterns on DSPS - delayed sleep phase syndrome fatigue in women with HIV/AIDS. J Assoc Nurses AIDS Care 2001; 12 Suppl:19-27. 202. McCarten JR, Kovera C, Maddox MK, Cleary JP. Triazolam in Alzheimer’s disease: desats - desaturation events pilot study on sleep and memory effects. Pharmacol Biochem Behav 1995; 52(2):447- DLMO - Dim Light Melatonin Onset 452. dom - dominant 203. Miller LG, Kraft IA. Application of actigraphy in the clinical setting: use in children with attention-deficit hyperactivity disorder. Pharmacotherapy 1994; 14(2):219-223. dur - duration 204. Paul MA, Pigeau RA, Weinberg H. CC130 pilot fatigue during re-supply missions to Dx - diagnosis former Yugoslavia. Aviat Space Environ Med 2001; 72(11):965-973. ECG - electrocardiogram 205. Redeker NS, Mason DJ, Wykpisz E, Glica B, Miner C. First postoperative week activi- EDS - excessive daytime sleepiness ty patterns and recovery in women after coronary bypass surgery. Nurs Res 1994; 43(3):168-173. EMG - electromyogram 206. Sadeh A, Horowitz I, Wolach-Benodis L, Wolach B. Sleep and pulmonary function in epi - epinephrine children with well-controlled, stable asthma. Sleep 1998; 21(4):379-384. EPS – Simpson Angus Extrapyramidal Symptom Scale 207. Sadeh A, Mcguire J, Sachs H et al. Sleep and psychological characteristics of children on a psychiatric inpatient unit. J Am Acad Child Adolesc Psychiatry 1995; 34(6):813- F- female 819. f/u - follow-up 208. Schlesinger I, Hering-Hanit R, Dagan Y. Sleep disturbances after whiplash injury: objec- g-grams tive and subjective findings. Headache 2001; 41(6):586-589. GH - growth hormone 209. Schnelle JF, Alessi CA, Al-Samarrai NR, Fricker RD, Ouslander JG. The nursing home at night: effects of an intervention on noise, light and sleep. J Am Geriatr Soc 1999; Grp - group 47:430-438. h/o- history of 210. Schweitzer PK, Muehlbach MJ, Walsh JK. Sleepiness and performance during three-day HADS – Hospital Anxiety and Depression Scale administration of cetirizine or diphenhydramine. J Allergy Clin Immunol 1994; 94(4):716-724. HAMD - Hamilton Depression Scale 211. Shamir E, Laudon M, Barak Y et al. Melatonin improves sleep quality of patients with HR - heart rate chronic schizophrenia. J Clin Psychiatry 2000; 65(5):373-377. h – hour(s) 212. Shilo L, Dagan Y, Smorjik Y et al. Patients in the intensive care unit suffer from severe IS: interdaily stability lack of sleep associated with loss of normal melatonin secretion pattern. Am J Med Sci 1999; 317(5):278-281. IV: interdaily variability 213. Shilo L, Dagan Y, Smorjik Y et al. Effect of melatonin on sleep quality of COPD inten- log - sleep log or diary sive care patients: a pilot study. Chronobiol Int 2000; 17(1):71-76. LSEQ - Leeds Sleep Evaluation Questionnaire 214. Shinkoda H, Matsumoto K, Park YM. Changes in sleep-wake cycle during the period from late pregnancy to puerperium identified through the wrist actigraph and sleep logs. Lt - left Psychiatry Clin Neurosci 1999; 53(2):133-135. M - male 215. Shirota A, Tamaki M, Tanaka H, Hayashi M, Shirakawa S, Hori T. Effects of volitional meds - medications lifestyle on rest-activity cycle in the aged. Psychiatry Clin Neurosci 1999; 53(2):271- mel – melatonin 272. 216. Stein MA, Blondis TA, Schnitzer ER et al. Methlyphenidate dosing: Twice daily versus MID – multi-infarct dementia three times daily. Pediatrics 1996;98(4):748-756. mins – minutes 217. Usui A, Ishizuka Y, Obinata I, Okado T, Fukuzawa H, Kanba S. Validity of sleep log MML – Mini-MotionLogger (AMI – Ambulatory Monitoring, Inc) compared with actigraphic sleep-wake state. Psychiatry Clin Neurosci 1998; 52(2):161- 163. MMDG - mild-moderately demented group 218. Usui A, Ishizuka Y, Obinata I, Okado T, Fukuzawa H, Kanba S. Validity of sleep log MMNDG - Moderate, Mild or No Dementia Group compared with actigraphic sleep-wake state II. Psychiatry Clin Neurosci 1999; MMSE - Mini-Mental State Exam 53(2):183-184. mo - month 219. Usui A, Ishizuka Y, Matsushita Y, Fukuzawa H, Kanba S. Bright light treatment for night-time insomnia and daytime sleepiness in elderly people: comparison with a short- MSLT - multiple sleep latency test acting hypnotic. Psychiatry Clin Neurosci 2000; 54(3):374-376. ndom -non dominant 220. van Londen L, Kerkhof GA, van den Berg F et al. Plasma arginine vasopressin and NINCDS-ADRDA – National Institute of Neurological and Commu- motor activity in major depression. Biol Psychiatry 1998; 43(3):196-204. 221. Van Der Meer YG, Swanink CM, Zitman FG et al. Randomised, double-blind, placebo- nicative Disorders and Stroke and the Alzheimer’s Disease and Related controlled study of fluoxetine in chronic fatigue syndrome. Lancet 1996; 347:858-861. Disorders Association 222. Wallace-Guy GM, Kripke DF, Jean-Louis G, Langer RD, Elliott JA, Tuunainen A. NH - nursing home Evening light exposure: implications for sleep and depression. J Am Geriatr Soc 2002; norepi - norepinephrine 50(4):738-739. 223. Wolter TD, Hauri PJ, Schroeder DR et al. Effects of 24-hr nicotine replacement on sleep NS - not stated and daytime activity during smoking cessation. Prev Med 1996; 25(5):601-610. Nt - night 224. Yesavage JA, Taylor JL, Kraemer H, Noda A, Friedman L, Tinklenberg JR. Sleep/Wake num - number cycle disturbance in Alzheimer’s disease: how much is due to an inherent trait? Int Psy- chogeriatr 2002; 14(1):73-81. OBJ - objective 225. Zhdanova IV, Wurtman RJ, Wagstaff J. Effects of a low dose of melatonin on sleep in OCP - oral contraceptive children with Angelman syndrome. J Pediatr Endocrinol Metab 1999; 12(1):57-67. OSAS - obstructive sleep apnea syndrome PLM - periodic limb movements APPENDIX A. ABBREVIATIONS PLMD/RLS - periodic limb movement disorder/restless legs syndrome ACT - actigraphy POMS-Profile of Mood States ADAS - Actigraph Data Analysis Software PRMT - Probed recall memory test ADHD - Attention deficit hyper-activity disorder PSG - polysomnography AIMS – Abnormal Involuntary Movement Scale PSQI - Pittsburgh Sleep Quality Index ANOVA - analysis of variance PTSD – Post Traumatic Stress Disorder ASA – Actigraphic Scoring Analysis pts - patients BL - baseline PVT - psychomotor vigilance task

SLEEP, Vol. 26, No. 3, 2003 358 Actigraphy Review Paper—Ancoli-Israel et al QOL - quality of life QWB - Quality of Well Being r - correlation coefficient RA- relative amplitude RCT - randomized controlled trial RDI – respiratory disturbance index Rt- right Rx - drug SAD – Seasonal Affective Disorder SANS – Scale for the Assessment of Negative Symptoms SAPS – Scale for the Assessment of Positive Symptoms SB - single-blinded SBJ - subjective SDAT – Senile Dementia of Alzheimer’s type secs - seconds SDG - Severe Dementia Group, Sdur - Sleep duration SE - sleep efficiency SEI – Sleep Efficiency Index SH - sleep hygiene SHAPS-D - Snaith-Hamilton-Pleasure-Scale SL - sleep latency SOff - sleep offset SOL - sleep onset latency SOn -sleep onset SP-sleep period SQ - sleep quality SSS - Stanford Sleepiness Scale STAI- State Trait Anxiety Inventory S/W - sleep/wake SWT - sleep wake transitions temp - temperature TENS - transcutaneous electrical nerve stimulation TIB - time in bed TST - total sleep time TWT - total wake time Tx - treatment VAS - visual analogue scale W - wake w/ - with w/o - without w/i - within WASO - wake after sleep onset wk - week yr - year 6SMT - 6-sulphatoxymelatonin

Companies AMI - Ambulatory Monitoring, Inc. MM - Mini Mitter

SLEEP, Vol. 26, No. 3, 2003 359 Actigraphy Review Paper—Ancoli-Israel et al Correlation of ACTCorrelation of (sum) w/ PSG for % sleep lowered by 2 outliers ACT showed more EEGs sleep than did PSG. in this population different making sleep scoring on PSG No inferential stats difficult; ACTfor compared to observations ACT appeared to detect TST expected reduction of induced by forced desynchrony protocol (poorer sleep when out of phase w/ temp ACTrhythm), suggests can detect a disturbance induced but by circadian pacemaker, no reference standard. ACT good for measuring sleep but not good in detect- WASO; SOL, W, ing quiet focus of report was validating BRM. Much of the data graphed thus not possible to give mean and S.D. There is an advantage to ACTusing when doing ambulatory BP monitoring to determine dipping status Comments from Reviewer from Comments ACT useful in children & adolescents Observation comparisons to 87%, Sensitivity ACT, ACT Specificity 90%. PSG; agreement rates (0.33 to 0.85 for max; 0.29 to 0.95 for sum) but not meaningful for this population. PSG compari- ACTson (r values): MAX TSTTWT 0.81, % sleep 0.78, 0.67; ACT SUM TST 0.91, % TWTsleep 0.61, 0.75. Movement freq. (weakly) predicts SBJ SE, SQ; Circadian time of going to sleep affect- ACTed TST & SBJ SE & quality (all reduced when starting sleep between 1000h and 1300h), but did not affect movement index or freq. ACT tends to represent quiet WACT as sleep; consistently underestimated SOL; No usable inferential stats for or Specificity, Sensitivity, Accuracy ambu- In patients undergoing latory BP monitoring compli- ACTance with (91%) superi- or to log completion (71%) (more so in children and in clas- young adults). No diff sification of hypertension sta- ACTtus with compared to log in 38 subjects completing both but dipper status (= greater than 10% drop in mean BP at Nt) was 55% misclassified using log compared ACTto sleep data (32% false negative and 23% false positive) paper ACT Loss of up to 28% of adoles- data from children & cents should be expected; 5 record for 7Nts to get ACT usable. Generally 5Nts yielded reliable results for Wsleep start, mins, SE, SP & sleep mins may require more than 7Nts of data to be reliable. Good reliability for sleep start time but poorest for SP. Severe dementia (MMSE<20) / NS / Small sample size no sleep disorders/ Healthy, NS / low num of subjects, all not blind to time of D F, NS / Meds, sleep disorders / Low sample size, forced wakes during night cessful ambulatory BP monitoring / NS / NS NS / Major genetic medical, prob- psychological or sleep affect lems; meds that could health sleeplessness; mental primary or sleep problems in relatives / Few inferential statistical analyses presented PSG: TST, Wtotal PSG: TST, time, sleep total W TST, ACT: %; time, ACTsleep % for both max. & observations - Staff ACT; sum S/W detection SOff, WASO, SOL, Log: BT, SOn, SOff, ACT: SBJ SQ; movement index, freq. TST, of movement onsets SOL; BRM ACT: PSG: SOL; (light and tone conditions) SOL Ambulatory BP monitoring Pt in data base who had suc- Outcome Measures Measures Outcome / Bias Inclusion / Exclusion Conclusion from Study sleep sleep start, SP, ACT: Wmins, SE mins during SP, PSG; Observation by staff / PSG; Observation by staff Actillume / wrist / 1 min / 003g. / Cole et al. ‘92 where) / Gaewhiler / ndom wrist / 30 sec / NS / ACCORD software (UK) + Horne et al 1994 algorithm + custom algorithms Actillume / dom PSG; BRM / 3 Action arm / 1 min / NS / (modified); Other: 0 crossing; SOL Log / Mini logger / ndom Ambulatory wrist / NS / NS / Monitoring Comparison Measures / Measures Comparison apparatus / Actigraph / Epoch Length / Placement Sensitivity / Scoring SOn & log used to estimate others Observation by SOff; (mother for Grp1) / MML / L ndom ankle if under 36 mos, ASA wrist / 1 min / .05g / ; Other: zero crossing 10 (10) / 86.4±60 yrs / Severe All Pts wheelchair dementia; bound 9 (9) / NS (19-20) / Normal log; (temp reported else- 8 (8) / F 27±1.4, M 33±8.3 / Normal 62 (62) / (6-71) / essential or secondary hypertension or normotensive Sample Size (Completed Sample Size / Age (Range) / Mean Study) Medical Conditions (12-60 274 (224) / Grp1 NS mos); Grp2 12.9 (11.2- 14.4yr); Grp3 15.0 (14.0- 16.2yr) / Normal Validation / NH / Entire 24-h; Validation (overlapping) 1Nt ACT; 1D observations every Staff PSG, 30 mins from 1000L to 1930L ACT Compared w/ other techniques; Forced desynchrony / Laboratory (dormito- ry used as lab) / Entire 24-h; BL 8Nt sleep midnight-0800; Sleep Dep.1: 26h; Forced desynchrony: 17 cycles (27hr WD) sleep 9 h, 18 h; Sleep Dep.2: 26hr; Recovery sleep: ACT 9hr; continuous ACT compared w/ other techniques / Home / Nocturnal only for NS h; 1Ntt PSG+ACT+BRM, 1+ Nt timeout, 1Nt PSG+ACT+BRM (w/ 4 imposed awakenings during nights of recording) Retrospective / Home / 20-24 h; 1 D ambulatory BP moni- ACTtoring + + log Study Criteria Design / Criteria Design Study Protocol Location / Home / Nocturnal / Validation to BT only for 30 mins prior Study1: to 30 mins post BT; ACT, 7D Study2: ACT, 4-7Nt ACT 7Nt ACT, 7Nt Study3: Evidence levels for technology papers for technology levels Evidence Ancoli-Israel (26) 3C Blagrove (87) 3C Blood (23) 3C Eissa (174) 3C Table 2— Table AuthorCitation - / Level Evidence Acebo (36) 4C- a

SLEEP, Vol. 26, No.3, 2003 360 Actigraphy Review Paper—Ancoli-Israel et al Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Hauri (175) ACT compared w/ other tech- 25 (25) /Grp 1: 44.5 (22-65) / PSG, other / NS / NS / NS / ACT: SOn, TST; FINGER Grp 1: Insomnia; Grp 2;/ ACT not very good at meas- ACT was not focus of study niques / Laboratory / 25.5 (19-40) /Grp 1: NS / Sadah et al. 1989 SLEEP SWITCH; SOn TST Normal sleepers /NS / NS uring SOL; underestimates it but used as a comparison for 3C Nocturnal only; 1Nt Insomnia, Grp 2, normal “quite markedly”; ACT better the sleep switch device PSG+sleep switch+ACT sleepers than the sleep switch device & logs at TST compared to PSG but overestimates it

Jean-Louis (22) Validation / Laboratory / 20 (20) / 29.95±8.98 (21-53) / PSG; BT & morning ques- PSG: TST, SOL; ACT: TST, NS / No sleep path / The ADAS ACT scoring soft- Nocturnal; 1Nt ACT+PSG Normal tionnaire / Gaehwiler / dom SOL, WASO Awakenings after SOn count- ware correlates w/ PSG 2B wrist / 60 / 0.1g / ADAS; ed only if > 3 mins r=0.97 for TST w/ average Other: bandpass 0.25-3.0kg discrepancy 12 mins.

Jean-Louis (16) Validation; ACT compared w/ 46 (46) / Grp1 30±9.0, Grp2 PSG / Grp1 Gaehwiler, Grp2 PSG: TST; ACT: TST Grp1 Normal / NS / NS ACT is valid for assessing other techniques / Laboratory; 46.5±10.8 / Grp2 Insomnia (AMI) / Grp1 dom arm, Grp2 S/W; ACT is useful to assess 3C Home /; Grp I. Nocturnal; NS / NS / NS / ADAS S/W in insomniacs; ADAS is Entire 24 h 1Nt ACT+log in a valid software for ACT in home, 1Nt ACT & PSG in different types of apparatus lab; Grp 2. Nocturnal Entire 24 h 1 D ACT+log+SSS in home, 3Nt ACT+PSG in lab

Jean-Louis (176) Validation / Home / 148 (148) / Grp1 30.1±13.1, NS /GAEWILER / Grp1 dom log; ACT: TST, SE Index, NS / NS / NS No difference in actigraph For first night effect study, Nocturnal; Entire 24-h; Grp2 28±10, Grp3 34±10, arm, Grp2 1 old, 1 new on SOL, WASO, Activity level, placement and (dom vs. ndom would have been better to 4C-a Grp1 24h ACT+log; Grp2 Grp4 29± 14, Grp5 25.7±5.0 dom arm, Grp3-1 on each Activity counts>sleep thresh- wrist or wrist vs. ankle) and include third night of data 24h ACT+log; Grp3 1 Nt / Normal arm, Grp4-dom wrist, Grp5- old, num arousal periods>3 reliability (between new also ACT+log; Grp4 1 Nt dom wrist / NS / NS / ADAS mins instruments and new com- ACT+log; Grp5 3 or more Nt pared to old) Observed no ACT+log first night effect

361 Jean-Louis (177) ACT scoring software / 26 (26) / 46.4±10.8 / NS SSS / (AMI) / NS / 30sec/1 PSG: TST, SE; ACT: sleep Insomniac / NS / NS ACT:PSG in insomniacs; Arousals after SOn defined as Laboratory; Home / min / NS / ADAS; Other: threshold, WASO, TST, SE ACT “is an excellent tool for longer than 3 mins scored as 1A Nocturnal Entire 24-h; 1 wk zero-crossing unobtrusive documentation of W; several different r values ACT at home + SSS; 3 Nt S/W activity in individuals reported for same data PSG+ACT (only 1 night w/…insomnia” PSG+ACT analyzed)

Jean-Louis (46) Unblinded, nonrandomized, 273 (273 (40-64) / normal, diary/ Actillume (AMI) / 1 ACT: TST, SOL, SE index, community dwelling residents CR of illumination were sig- Higher amplitude of activity observational study, cross- NS min / NS / NS / Automatic mesor, amplitude (of the of San Diego identified by nificantly associated w/ activ- (by ACT) was associated w/ 4C-a sectional study / Home / scoring rhythm; ACTION3 cosine) and phase (timing of random telephone survey / ity and sleep rhythm meas- less reported daytime nap- Entire 24-h; 3D ACT the peak of the fitted cosine) NS / NS ures. Higher amplitude of log ping. (r=-.16, p<.05). Level of illumination illumination correlated with sleep phase (r = .16), lower SE index (r=-.15), and less reported daytime napping (r=-

Actigraphy Review Paper—Ancoli-Israel etal .16). Higher amplitude of activity correlated with sleep amplitude (r=.30), Sdur(r=.21) and less daytime napping (r = -.16). Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Jean-Louis (12) Calibration & Validation / Grp 1; 39 F (37) Grp 2: 4 M PSG / Actillume / wrist / 1 NS Healthy /NS/NS Grp 1: In bed: minute-by- Authors conclude that Grp 1: home; Grp 2, laborato- & 7 F (NS) / Grp 1: 63.7 (51 min / 0.003g / zero-crossing minute agreement rate ACT Actillume useful for deter- 3C ry/ Grp 1: 24h ACT + PSG; – 77) Grp 2: 25.4 (19-34) / threshold with Action 3 with PSG = 85%, r = 0.98; mining S & W at home & in Grp 2; 5Nt & 4D nocturnal Grp 1; post menopausal mean difference between PSG lab with elderly and young, PSG+24h ACT & ACT = 21 mins, sensitivity however different scaling fac- = 94.8%, Specificity 40.6%. tors in the scoring algorithm ACT missed much midsleep need for optimal results W; suggests may need sec- depending if ACT used in bed ond-by-second monitoring of only or over 24h and with activity to catch these. 24h: individual differences such as minute-by-minute agreement age and likelihood of rate ACT with PSG = 89%, r WASOs. Shows need for = 0.90; mean difference more research to optimize use between PSG & ACT = 1 of ACT in various situations min. with various individuals.

Grp 2: 24h using Webster et al., (1982) scoring rules: minute-by-minute agreement rate ACT with PSG = 91%, r = 0.92; mean difference between PSG & ACT = 5 mins.

Jean-Louis (24) Validation; ACT Compared 5 (5) / Grp1 25± 6 / No med- PSG / Actillume & MML / PSG: TST, SEI, W, S; ACT: No medical conditions, emo- Actillume performed compa- w/ other techniques; ical conditions, emotional ill- wrist / 1 min / 0.003g / min by min agreement w/ tional illness, or sleep distur- rably to MML and the differ- 2B Comparison of 2 ACTs & 5 ness or sleep disturbance Action3; Other: Actillume- PSG, % PSG W misscored as bances / NS / 5 subjects of ent modalities of each were quantification modalities / sum of activity and maximal sleep, % PSG sleep misscored unreported sex; data from 10 all comparable for S/W analy- 362 Laboratory / Diurnal; 5Nt & activity; M M-zero crossing, as W, sensitivity, specificity of the 25 nights were not use- sis. Although all 5 modalities 4D nocturnal PSG+2 ACTs time above threshold, & pro- able; no ANOVA to compare performed well, proportional portional integrating modali- devices or modalities integrating modality seems ties better overall for this age Grp and zero crossing mode least desirable

Actigraphy Review Paper—Ancoli-Israel etal at end of each Tx period

Krahn (30) ACT compared w/ other tech- 60 (30) / Median 58 (20-87) / NS / (AMI) / wrist / NS / NS Log: TST, WASO, TIB, SOL; Consecutive admits to med- Nurses observations of sleep ACT as comparison standard niques / out-of-lab hospital NS / Cole et al. ACT: S, W, TST, WASO, psych or adult acute care in psychiatric patient agrees for SL & nurse observation of 4C- a bed / Entire 24-h; 3D ACT & TIB, SOL; Observations by inpatient units / Patient seek- satisfactorily w/ ACT but pts patient log in A.M. nurses for S, W ing 1:1 nursing care or other self-reports using log in a.m. conditions / Patient selection is not satisfactory limited to those able to comply w/ study design Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Lockley (29) ACT compared w/ other tech- 49 (49) / 46.6±12.2 / Blind log; mel; Urine samples / log: SOL, SOn, SOff, Num Meds influencing sleep or ACT and logs yielded similar Unique use: to see if sleep niques / Home / Entire 24-h; Motion loggers or MML / awakenings, dur awakenings, mel / NS / NS results for some aspects of (daytime nap num & dur & 3C mean nights/subject=23±7 wrist / 1m / NS / Action 3; num naps, dur naps; ACT: sleep but not others. night sleep dur) greatly (range 6-35) of ACT + log Other: Zero crossing mode SOL, SOn, SOff, Num awak- Proportion of pts showing dif- altered by CR type (normal enings, dur awakenings, num ferences between ACT and entrained, abnormal naps, total nap dur log greater than those show- entrained, or free run) ing no differences.

Middelkoop (40) Validation / Home / Entire 10 (10) / 24.1±3.1 (19-33) / Log / Gahwiler / both wrists log: Time to bed, light out Rt banded, healthy, no meds / Activity recorded at wrists > 24-h; 45 h ACT NS & both ankles & trunk at time, SOL, num awakenings, NS / NS ankles > trunk. Motor activity 4C-a navel / 5 sec (min) / 0.1g / rise time, num naps, nap dur; lower at Nt comparably at all ACTSTAR 1.0; Other: 0.25- ACT: Activity level, epochs>, sites but significantly more 3kg dur of epochs>0 activity of dom wrist during diurnal period, & wrist correlated more w/ trunk at Nt than ankles; opposite diurnally. Ndom wrist activity correlates more w/ trunk activity than does dom wrist activity; During sleep ACT reveals more whole body, generalized movement but during W there are more isolated limb movements

363 ankle, trunk (navel) / 5 sec- dom wrist placement yielded ing level of waking activity. onds / NS / NS greater diurnal activity counts than ndom. All sites clearly showed circadian sleep-W differences. Suggests more study w/ placements concomitant w/ PSG

Minors (45) Technique Development / 8 (8) / Grp1 49, Grp2 (18-20), log / Gaehwiler / ndom wrist / NS NS / NS / Low sample size Method developed for com- Home / Entire 24-h; 3-26 24- Grp3 26, Grp3 (43-74) / 6 mins / NS / NS paring level of activity in bed 4C-b hperiods of ACT Normal Grps 1&2, CR disor- vs. out of bed (=”dichotomy der, DSPS Grp 3; Colorectal indices”); Healthy subjects cancer Grp 4 showed greater dichotomy than either pts w/ DSPS or colorectal cancer; Might be useful technique for study of Actigraphy Review Paper—Ancoli-Israel etal circadian phase shift (shift- workers, jet lag) & illness

Minors (96) Repeat measures, 2 conditions 8 (8) / (18-35) / Normal log / Gaehwiler / ndom wrist / ACT: relative activity levels Night shift nurse / NS / NS When data is corrected for ACT can be useful in study- compared / Home / Entire 24- 30 sec/1 min / NS / NS length of sleep D sleep has ing adjustment to shift work 4C-a h; 2 Ds rest ACT & 3 Nt greater activity levels than This paper is a further analy- shifts ACT night sleep; Difference sis of data in an earlier study. between W activity and sleep activity levels tends to be less during night work shifts Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Monk (31) ACT compared w/ other tech- 4 (4) / 42.5 (38-47) / Healthy PSG; log / Actillume / ndom PSG: SOn, SOff, Sdur, SE, No sleep disorders; 2 astro- ACT during space flight During space flight niques / Space Shuttle arm / 1 min / NS / 5 consecu- SOL, WASO, Sleep stages; nauts, 2 payload specialists / clearly identified SOn & SOff 3C Columbia / Entire 24-h; tive mins of 0 (sleep) or non- ACT: SOn, SOff, Sdur SE, NS / Only 4 subjects better than log as well as Diurnal; 72 h ACT+PSG+log; zero (W onset) counts SOL, WASO sleep dur & SE as compared 7D nothing; 72 h to PSG. Even good correlation ACT+PSG+log of ACT counts & different sleep stages, but ACT failed to detect a .98 min SOL. Overall ACT is a simple efficient means of evaluating sleep in space when PSG is not feasible

Otsuka (178) Validation / Laboratory; Out- 44 (44) / Grp1 (28-46), Grp2 ECG; BP; HR / Activetracer / ACT: Activity level; BP Simultaneous ACT & BP can Hard to pull data out of this of-lab hospital bed / Entire (28-78) / Normal; Healthy waist / 1 sec / 0.01 to 0.50g / be used to show relationship article; uses a lot of individ- 5D-a 24-h; 72+h ACT 7 ECG, (also NS of S/W w/ BP. Suggests S/W ual examples. BP for 48h & HR for 24h) rhythms of about 7 Ds (cir- caseptan) in addition to circadian & a 3.5 D (circasemisep- tan) rhythm in irregular S/W cycles but able to identify individual differences in all these

Pollak (18) Validation; 2 ACT devices 5 (5) / 33.20 (22-47) / Normal None / 1. MML (AMI Model ACT: Activity counts NS / NS / Low sample size; Both devices were reliable. Good study; almost pilot compared reliability, validity, 20.000), 2. Activity monitor wide range of durs of ACT Related is the need for the study; showing need for stan- 4C-b artifact rejection / NS / 7-92h (Gaehwiler Electronics model measurement; no comparison instrument to detect small dardization of criteria for ACT (mean=41.5) 280 32k v1) / ndom wrist / 30 measures sleep movements and inter- acceptability in ACT instru-

364 sec / for Mini-Motionlogger- ruptions; again G was poor. ments manufacture’s software for Because of above, G would Gaehwiler-NS over estimate sleep and sleep continuity. AM is better than G for S/W detection Pollak (21) Validation & Reliability / 15 (14) / NS (20 – 35 & 70 – PSG / ActiTrac & CSA NS Healthy with no sleep prob- ACT is adequate for determi- Unique scoring and data pro- Laboratory / 7 D + N ACT + 72) / healthy, no sleep prob- Activity Monitor / non-dom lems nation of S/W circadian cessing. Makes a strong case 3C PSG in time isolated, free-run lems wrist / 30 secs / 0.024 & rhythm, rest/activity cycles, for superiority of PSG over condition 0.033 g / 1) single threshold and S-consolidation but not ACT for most applications. & 2) “logistic regression” otherwise as a substitute for with moving 20 minute win- PSG because it overestimates dow S dur and SE but misses WASOs. No differences between the two ACT apparatus. Sensitivity = 86.6% for N Actigraphy Review Paper—Ancoli-Israel etal only, 62.2% N + D because failed to show D S. Specificity = 89.6% for N using logistic regression; approached 100% for D. Agreement rate (a poor measure in general) = 76.9% for D + N, 82.0% for N only. ACT & PSG have poor agreement on W to S and S to W transitions. Some indication that the stages of sleep are reflected in level of ACT ranging from high for W to stage 1 then REMS then stage 2 then SWS at low end. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Reid (25) Validation; ACT compared w/ 32 (32) / Grp1 21.2±2.7, Grp2 PSG / Gaehwiler / ndom hand PSG: Sleep, Wake, SOn, NS / Smokers, taking meds, ACT valid for S/W activity & other techniques / Laboratory 43.9±6.8 / NS / 30 sec / 0.1g 125 ms sam- SOff; ACT: S, W, SOn, SOff sleep disorders / Lab shift sleep dur (although ACT dif- 3C / Diurnal; Entire 24-h; pling time (8 kg) Baudpass work studies may not com- ferent from PSG for SOff) but ACT+PSG (when sleep) for filter 0.25-3.0 kg / NS pletely reflect real world, not as good for more specific 1Nt adaptation, 2D of 12 h D their older subjects were real- measures such as SE; ACT: shift (night sleep), 2D 12hr ly middle-aged. agrees more with PSG for night shift (D sleep), 1 1/2D young adult than for middle- “off” aged because more quiet W scored as sleep in latter , as the likelihood of sleep decreases so does ACT accuracy.

Roehrs (37) Validation: ACT compared w/ 34 (17) / 27.9±4.3 (19-35) / MSLT / MML / ndom wrist / PSG: MSLT, SOL; ACT: Healthy, normal S/W patterns, ACT can be used to show ACT used to measure daytime other techniques / Laboratory Normal 30 / NS / Action-W SOL, % inactivity, counts per no sleep complaints, <250 changes in daytime epochs of sleepiness (lapses) 2B / Diurnal only for 11 h; 1D active epoch mg/D caffeine consumption, inactivity due to changes in placebo-8hr TIB, 2-7D recov- no Rx of drug/alcohol abuse, sleepiness. MSLT show dif- ery, 1D placebo-4hr TIB, 2- non-smokers / NS / ACT ference between 8h, 4h, 0h 7D recovery; 1D placebo-0hr scoring software not specifi- while diphenhydramine inter- TIB, 2-7D recovery, 1D cally developed to assess day- mediate to & differed from diphenhydramine 50mg-8hr time sleepiness both placebo-4hr TIB & TIB; in Latin Square design placebo-0hr TIB. Counts during active epochs didn’t yield any significant difference. MSLT was more sensitive than ACT to sleep loss but ACT may represent more real-world index of effects of 365 sleepiness

Sadeh (35) Comparative Tx; ACT com- 50 (50) / 14.11±4.21 (9-24) / log / (AMI) / NS / NS / NS / log: SOn, Sdur, Sleep%, night Nt waking problems & whose ACT+log showed improve- ACT+log showed improve- pared w/ other techniques / Nt waking problems ASA awakenings; ACT: SOn, Sdur, parents complied w/ an inter- ment during time a Tx was ment during time a Tx was 4C Home / 1 wk ACT+log BL; 1 Sleep%, night awakenings, vention / NS / ACT scoring applied but no difference applied but no difference wk or more ACT+log during quiet sleep, active sleep software not specifically between Txs; Parental log not between Txs 1 of 2 interventions developed to assess daytime as good as ACT re: whether sleepiness, No untreated con- children’s sleep improved trol Grp during Tx (60% agreement rate)

Sadeh (33) ACT compared w/ other tech- 66 (66) / 14.7±4.6M (7-26M) log / AM-16 / leg / NS / NS / log: SOn, Sdur, Sleep percent, Nt waking problems / NS / Both ACT+log should be Paper rightly concludes that niques / Home / Nocturnal / Healthy ASA num awakenings; ACT: SOn, pts all referred to sleep clinic used in assessing sleep prob- both ACT+log should be used 4C-a only for NS h. mean=6D Sdur, S%, num awakenings for night waking problems lems of infants in assessing sleep problems of (range: 3-12) ACT+log infants because either alone

Actigraphy Review Paper—Ancoli-Israel etal distort or miss some data Sadeh (17) Validation; ACT compared w/ 44 (44) / Grp1 22.6±1.7 (20- PSG / AMA-32 / 1. 1 dom & PSG & ACT: S, W - Minute Normal / NS / NS S/W derived from ACT is Paper contains suggestions other techniques; reliability / 25), Grp2 13.8±1.9 (10-16), 1 ndom wrist, 2. 2 on same by minute comparison robust & little affected by for proper use of ACT 1A Laboratory & Home NS / Grp3 (12-46) / NS wrist / 1 min / NS / zero device placement although Nocturnal; 1. PSG & ACT; 2. crossing (& based on 1st 10 activity levels differed 42-48h w/ 2 ACT on same adult pts) between dom & ndom wrist wrist during SP & wakefulness but resulting ability to discriminate sleep was relatively unaffected by this. Sforza (39) validation / laboratory / noc- 35 (35) / 54.8±1.6 (37-72) / PSG; Anterior tibialis EMG / PSG: PLM (via EMG); ACT: NS/Technical problems dur- ACT: EMG Movements – leg Conclusions differ from those turnal only for 7.5 h, 2N (s/b 13 OSAS, 22 EDS or snoring Gaehwiler / upper Rt foot / 5 PLM ing recording/ Most pts Dx ACT not valid. R=0.78 but of Kazenwadel, et al., 1995 2B Nt? PSG/ACT(n= 10; CPAP or RLS/PLMD sec / 0.1g / NS w/O SA ACT underestimates move- due to more sophisticated titration on night 2) in ments, especially those <3 sec analysis PSG/ACT (n=25) and < 50 mv but was reliable across 2 nights in 8/10 subjects thus not valid for DX PLMD but valid for Tx evaluation. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Shinkoda (27) ACT compared w/ other tech- 4 (4) / (20-33) / NS PSG / AMA-32CL / ndom PSG: S,W, TST, SOL, SEI, NS / NS / very low sample ACT correlates w/ PSG for niques / Laboratory / wrist / 30s (but 1 min com- WASO, num awakenings; size TST, SOL, SEI, WASO, num 3C Nocturnal; 3Nt ACT+PSG pared to PSG) / NS / Cole et ACT: S, W, TST, SOL, SEI, awakenings al 1992; Other: Zero crossing WASO, num awakenings mode Tikotzky (34) Observational /NS /put on 1 h NS (59) M, 29, F, 30 ( 3.8- Sleep log-diary / AMA-32; ACT: SO, TST, % sleep, # NS/NS Parents log correlated well before sleep, removed 1 h 6.1) /normal AMI / non-dom wrist / NS / wakes with ACT but less accurate in 4C-a after wake, 4-5 N NS / Sadeh ASA sleep quality measures. 41% of kids had fragmented sleep

Van Someren (179) Technical considerations / 25 (25) / Grp1 28±4, Grp2 NS / Entran / dom wrist / NS NS NS / NS / NS Use of bandpass of 0.25 to Laboratory; while performing 80±9 / NS / ±2g / NS 11kg for ACT is better to 4C-b 6 tasks / NS measure all movement. May under record movements especially in young. It also tends to reduce “gravitational artifact”.

Van Someren (14) Technical / NS / NS 25 (25) / Grp1 Young Adults, None / NS NS NS / NS / ACT scoring soft- The bandpass filter of typical Highly theoretical w/ little Grp2 Elderly / NS ware not specifically devel- ACT appears to be too narrow information on where the data 5D-a oped to assess daytime sleepi- & may miss some movements shown came from ness .5-11Hz may be better. This is especially important for studies of aging since younger tend to have more rapid movements that can be missed w/ the prevailing cut- offs 366 Verbeek (180) Validation / laboratory / NS 20 (20) / 47.7 (33 – 64) / PSG; log / NS / NS / NS / NS healthy / NS / NS ACT overestimates TST in Too many procedural details insomnia NS / Cole et al., 1992 insomniacs by mistaking not specified. 3C lying still but awake for sleep. PSG better than ACT; both better than logs.

Violani (42) Parallel design / Laboratory 16 (16) / (19-28) / NS PSG; log / 4 motion logger / ACT: Activity levels Normal sleep dur and sched- During W Rt hand shows (nocturnal study); Home each wrist & each ankle / 1 ule, healthy / NS / NS more intense activity than Lt 4C-a (diurnal study) / Entire 24-h; min / NS / zero crossing that disappears during early 1wk log; 1Nt PSG+ACT on Action 2 sleep but reappears during each wrist & each ankle, 1Nt later sleep that cannot be PSG+ACT on either wrist or accounted for by a difference each ankle, 1Nt PSG+ACT in in arousal. No such differ- opposite positions from previ- ences noted in ankle activity.

Actigraphy Review Paper—Ancoli-Israel etal ous night ACT & respiratory variables provided good assessment of RDI and was somewhat better than resp variables alone Comments from Reviewer from Comments No validation of ACTNo validation of other ACTthan pt. report used only to aid initial Dx of DSPS. Outcome assessed only by ACT. questionnaire, not Tx improved: num PLM / h, TIB w/o leg (p,.001) % of movements, (p<.001) SBJ QOL, (p<.001) num of PLM episodes, (p<001) num of PLMs per episode, (p<.001) dur of PLM episodes (p=.005) ACTr=.74, p<.0001; for RDI & PSG RDI, r=.976, p<.0001; TIB, r = .94 for RDI based on Sensitivity & negative predic- TIB RDI tive value lower for ACTthan RDI in severe OSA patients. No sig. diffs, in ACT in No sig. diffs, sleep variable b/t PTSD pts & normal controls; no correlation ACTb/t sleep variables & AM reports; no correlation b/t ACT sleep variables & sleep parameters on health questionnaire No significant correlation ACTbetween & apnea, EEG arousal, or respiratory phase; L ankle ACT/L tibia EMG r=.73 (p<.001); L ankle ACT/L wrist ACT r=.55 (p<.001); detection of limb movement predictive of arousal, but EMG more than ACT paper ACT sleep times related to symptoms; teacher-reported to decreased sleep related more externalizing symptoms Acrophase of 6-SMT advanced by bright light, no Advance at f/u. difference sustained only in late-phase pts. Late-phase pts SOn and advanced by bright SOff light. 96.7 % report mel helped DSPS, 91.5% relapse after mel cessation 2 wakes ≥ 1 yr prior to study ≥ longer than 20 mins or 1 longer than 1 h, Increase in PLMs 2nd half of night / NS / None Improvement in Symptoms w/ L-dopa 1st 4 h, Rxs Free of sleep-affecting for 10-Ds / NS / NS Tx; only 1 pts presenting for Nt assessed No meds/no meds / small high sample size, low power, I error due to multiple Type tests NS / non DSPS, serious health problems, regular med use, Rx/alcohol abuse, serious psychiatric problems, recent travel/shift work / NS finished / No control Grp ACT: PLM per h of TIB, time PLM per h of ACT: w/o movement; Others: QOL ratings; physician-rated clinical global impressions, SBJ QOL log; PSG; Health questionnaire Epworth ACT; PSG; NS / NS / Sample limited to ACTPSG sleep variables, NS/NS ACT For TSTTST, & PSG Outcome Measures Measures Outcome / Bias Inclusion / Exclusion Conclusion from Study Behavioral: child behavior form checklist, teacher report behavioral; Mood; ACT; Log; 6-SMT Survey Tx, Dx, mel No meds, DSPS Questionnaire; Observation by others (physician) / Movoport (Rimkus) / foot / NS / NS / NS log / (AMI) / wrist / NS / NS / Sadeh et al ‘89 PSG; Epworth / IC Sensors / L wrist, L ankle / NS / NS / Custom Resp variables / PSG, Actiwatch / non- Cambridge dom wrist / 1 min / NS / NS Comparison Measures / Measures Comparison apparatus / Actigraph / Epoch Length / Placement Sensitivity / Scoring AMI mini-motion log- None / AMI / ger / belt / 1 min / NS / NS / NS None / MML (AMI) / wrist / NS / NS / NS Sadeh; Other: zero-crossing mode 37 (30) / 58±10 / PLMD/RLS 27 (27) / Grp1 31.9±6.31, Grp2 30.5±5.5 / normal; PTSD 40 (36) / 48±12 (20-73) / Snoring, EDS 5 / 52 (15) / 20 (20) M, 15, F, OSAS Sample Size (Completed Sample Size / Age (Range) / Mean Study) Medical Conditions 23 /9.73 49 (NS) M, 26, F, (1.39) / normal 78 (45) / 25±6 / DSPS Actillume / wrist / NS None / 61 (NS) / 30.17±11.26 (16- 61 (NS) / 30.17±11.26 54) / DSPS 1 yr after 6-wk mel ≥ Controlled clinical trial, cross-over design / Home / Nocturnal; 5Nt BL 5Nt ACT; ACT placebo 5Nt ACT; Rx Case series / Home / Entire ACT24-h; 4-7D BL to help diagnose DSPS; f/u questionnaire Cohort study / Home / Nocturnal 8 h ACT compared w/ other techniques / Home / Nocturnal only for 6 h; 1Nt PSG/ACT ACT compared with other techniques /lab / nocturnal ACT only for 8 h 1 Nt PSG, Study Criteria Design / Criteria Design Study Protocol Location / / home / Observational study ACT entire 24 h, 3D parallel / Home / Entire RCT, Tx; 2D post 24 h 2D BL; 5D Tx Tx Evidence levels for studies of sleep disorders sleep of for studies levels Evidence Collado-Seidel (73) 4C-b Dagan (57) 5D-a Dagan (54) 5D-b Drinnan (70) 5D-b Elbaz (71) 3C Table 3— Table AuthorCitation - / Level Evidence Aronen (67) 4C-b Cole (44) 2B

SLEEP, Vol. 26, No. 3, 2003 367 Actigraphy Review Paper—Ancoli-Israel et al Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Franck (62) Cohort study / Home / Entire 33 (33) / Grp1 10.61±3.5, log; Self report; Parental log; ACT NS / For controls: Chronic Parental reports indicated 24-h; 3D ACT Grp2 10.87±3.27 / Normal, report / MML (AAM-32) / medical conditions, regular HIV-infected children to have 4C-b HIV+ NS / 2 mins / NS / Action 3 med use / All infected pt’s less sleep, more awakenings, undergoing Rx therapy & more nightmares than controls; child logs showed no diff. in sleep characteristics or daytime fatigue; ACT showed greater num of awakenings, lower SE, W time during the Nt, less SE for HIV+ pts

Guilleminault (51) NS/ Home / Nocturnal; 1Nt 32 (30) / Grp1 44±8, Grp2 Questionnaire; log / MML / log; ACT; Self-reported sleep: SL > 30min or TST < No diff for SH or SH + TST from logs < ACT TST, PSG screening, 4wk Tx; 4Nt 44±8, Grp3 44±8 / Insomnia dom arm / 2 mins / NS / NS TST, SL, num W 360min, No Rx sedative-hyp- Exercise, for SH + SL logs > SL ACT, W logs < 4C-b ACT; 7Nt log notic use / NS / NS light, TST higher, SL lower, W ACT W lower

Guilleminault (181) Case series / Laboratory and 184 (184) / M 34±8 (18-52), PSG, MSLT, Epworth, SSS, Log; PSG: MSLT, AHI, NS / NS / self-selected sam- EDS associated w/impaired ACT had little role in results home / Nocturnal only for 8 F 39±12.5 (18-54) / log / NS / NS / NS / NS / NS SOREMPs; Behavioral vari- ple daytime function. 24-h coma, or conclusions. 5D-b h; 1 Nt PSG; 7 Nt log and Head/Neck trauma and hyper- ables; Mood variables head fracture, or immediate ACT somnia neurosurgical intervention associated w/ ESS> 16 and MSLT <5. Sleep apnea associated w/ whiplash.

Hayes (182) Cross-sectional study / Home 30 (27) / Grp1 80.9mo±7.2 Time-lapse video / ACT: AFFT Normal sleeper; mid-infancy Movement intensity & move- / Nocturnal only for 9 h; 1Nt SEM / Normal Electrostatic bed pad / to pre-puberty / NS / None ment bout structure integrity 4C-b ACT head/shoulder & buttocks / 1 during sleep decreased w/ sec / NS / FFT, Periodgram age; decline in bout dur w/ age; increase in bout freq. w/ 368 age

Hering (63) NS / Home / Entire 24-h; 3D NS (NS) / Grp1 8.0±3.0, log /(AMI) / wrist / NS / NS / log; ACT NS / NS / NS log of parents showed differ- ACT Grp2 8.0±2.3 / Normal, Sadeh ‘89 ences in AM W time, early Nt 5D-b Insomnia, Autism arousals, & multiple Nt arousals

Jean-Louis (53) Observational study / Home / 273 (NS) / 51 (40-64) / None / Actillume / wrist / 1 Mood: QWB scale, CES-D NS / NS / Mean age some- ACT-measured sleep vari- Entire 24-h; 3D ACT Normal min / NS / Action 3 what high (N50) to be repre- ables not significantly corre- 5D-b sentative of the adult popula- lated w/ QWB or CES-D tion; used pts received from scores random telephone dialing- those who consented may be more likely to have problems

Actigraphy Review Paper—Ancoli-Israel etal Jean-Louis (16) Validation; ACT compared w/ 46 (46) / Grp1 30±9.0, Grp2 PSG / Grp1 Gaehwiler, Grp2 PSG: TST; ACT: TST Grp1 Normal / NS / NS ACT is valid for assessing other techniques / Laboratory; 46.5±10.8 / Grp2 Insomnia (AMI) / Grp1 dom arm, Grp2 S/W; ACT is useful to assess 3C Home /; Grp 1 Nocturnal; NS / NS / NS / ADAS S/W in insomniacs; ADAS is Entire 24 h 1Nt ACT+log in a valid software for ACT in home, 1Nt ACT & PSG in different types of apparatus lab; Grp II. Nocturnal; Entire 24 h 1 D ACT+log+SSS in home, 3Nt ACT+PSG in lab

Jean-Louis (177) ACT scoring software / 26 (26) / 46.4±10.8 / NS SSS / (AMI) / NS / 30sec/1 PSG: TST, SE; ACT: sleep Insomniac / NS / NS ACT:PSG in insomniacs; Arousals after SOn defined as Laboratory; Home / min / NS / ADAS; Other: threshold, WASO, TST, SE ACT “is an excellent tool for longer than 3 mins scored as 1A Nocturnal Entire 24-h; 1 wk zero-crossing unobtrusive documentation of W; several different r values ACT at home + SSS; 3 Nt S/W activity in individuals reported for same data PSG+ACT (only 1 night w/…insomnia” PSG+ACT analyzed) Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Kazenwadel (38) Validation / Laboratory / 30 (29) / 51.0 (29-74) / PSG / MOVOPORT / top of PSG: PLM, TIB, SEI; ACT: RLS + PLM / NS / NS Measurement of PLMD by Method requires considerable Nocturnal; 2wk BL, 4wks PLMD/RLS Rt foot between 1st & 2nd PLM, TIB ACT is possible w/ 1/2-sec- manual adjustment of com- 1A levodopa & lenserazide, 4wks toes on pts reported most ond epochs. Med made no puter analyzed ACT data to placebo in randomized affected leg / 0.5 sec / NS / difference in EMG & ACT avoid underestimation. crossover design; 2Nt computer dichotomizing rela- correlations Movements of only one leg ACT+PSG w/ EMG at end of tive to a threshold followed recorded by ACT BL (1st was adaptation night) by manual modifications then 1Nt ACT+PSG w/ EMG at end of each Tx period Kerkhof (56) Correlational / Home / 14D 80 (80) / 34.8±10.1 / log; Temp / Gaehwiler / ndom log; Temp; ACT: mean activi- 6 mos. Serious sleep com- Significant difference in MR oral temp, VAS SBJ alertness; Insomnia hand / 30 sec / 0.1g / custom ty count (MR), immobility plaints / NS / NS b/t early & late temp phase 4C-b 14Nt log, VAS SQ.; 11D ACT (IMM), fragmentation (FR) pts no difference for FR or IMM q(p) values greater for early phase Grp

Kramer (68) Cross-sectional study/ home / 40 (40) Grp 1 M 21. Grp 2 Sleep log, diary/ Gaewhiler / Sleep diary, log; ACT Exclusion: psychiatric, neuro- Young subjects had more TIB 14 D sleep logs, 11 D ACT, M, 19 / Grp 1 65.1 (59-74 non-dom wrist / 30 sec / 0.1g logical disease, cardiovascular variability than elderly sub- 4C-b 14 D oral temp (4.4) Grp 2 20.8 18-26 (2.2) / / NS disease, drug abuse / NS jects; p<.001. Elderly sub- normal jects had advanced sleep phase relative to younger subjects. Elderly subjects had more midnight awakenings than young subjects. Sleep logs showed less TST and SE for elderly subjects.

Kunz (183) Unblinded, nonrandomized / 9 (9) / 57 (40-71) / PLMD, PSG, log / 2AK / wrist / NS / Log; PSG; ACT NS / RLS / small sample Tx improved well being in 7 No measurement of sleep Laboratory, Home / Nocturnal Depression, RSBD, NS / NS of 9 pts, PSG movement variables w/ ACT 369 4C-b only 8 h; 2 Nt PSG BL; 14 Nt Parkinson’s dysautonomia parameters improved w/ Tx, ACT BL; 2 Nt PSG post Tx; ACT measured movement 14 Nt Tx reduced by Tx Kushida (19) Validation / Laboratory / 100 (100) / 49±14.7 / OSAS, PSG; log / AW4 MM / ndom log; PSG; ACT NS / NS / self-referred pts, ACT over estimated TST by Nocturnal 8 h 1Nt PSG, ACT, narcolepsy, insomnia, wrist / 30s / .01 g / Oakley mixture of Dx 1.0-1.8 h; log overestimated 2B log PLMD/RLS, UARS ‘97 TST by 0.3 h; num of awakening more accurate for ACT than log; ACT high threshold algorithm: sensitivity = .98; specifying = .76 accuracy = .28

Middelkoop (69) Validation / Home / 1Nt ACT, 167 (116) / 53.6±10.7 / OSAS log; Oronasal thermistry / Log; ACT; Respiration (apnea Habitual snoring w/ excessive AI ≥ 5 Grp. Greater than AI log Gaehwiler / ndom arm / 15 index) daytime sleepiness and/or <1 & AI < 5 groups for

Actigraphy Review Paper—Ancoli-Israel etal 4C-b sec / NS / % epochs count>0; nocturnal respiratory arrests / movement index & frequency Other: Counts/epoch, dur of NS / Apnea of pts was mild, Index. DIP sig diff in 3 Grps, uninterrupted activity, dur of may have led to lower dis- ETOH related to DIP, DIP immobility (DIP), fragmenta- crimination correlated w/ AI tion index

Nagtegaal (61) Controlled clinical trial, DB, 30 (25) / 37.3±15.3 / CR dis- PSG; log; Temp / NS / NS / Log; temp; PSG; ACT NS / Under age 12, prior use Temp markers showed cross-over design / order NS / NS / NS of mel, liver disease, renal advance after Tx; pts more 1A Laboratory, Home / failure, severe neurological or refreshed (logs) w/ Tx com- Nocturnal, Entire 24-h; 1D psychiatric disorder, pregnan- pared to placebo rect temp (pre Tx), 1Nt PSG cy / NS (Placebo mel), 3D ACT (placebo), 14D ACT (mel), 1D rect temp (post Tx) Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Okawa (58) Comparative Tx / Home / 11 (11) / Grp1 32.3 (16-46), Log; temp / (AMI) / ndom Log; temp; ACT NS / NS / Small sample size 6/11 pts responded to mel Tx; No statistical analysis report- Entire 24-h / 14D ACT, body Grp2 28.4 (17-34) / CR disor- wrist / NS / NS / NS temp phase shifted first, fol- ed, no controls 4C-b temp BL, Tx (time not indi- der lowed by shift in S/W cycle cated)

Sadeh (66) Cross-sectional study / home / 140 (140) M 72, F 68 Grp 1 Questionnaire, sleep log/ Sleep diary/log, ACT Exclusion: no acute physical Night –to-night reliability for 4-5 D ACT on school nights; N=50; Grp 2 N= 37; Grp 3 N diary / AMI AMA-32 / non- illness or behavioral problems ACT > .70 for most measures. 4C-b sleep logs, sleep habits ques- = 53 / grp 1 7.0 (7.2-8.6 dom wrist / 1 min / NS / ASA Older subjects had delayed tionnaire (o.34); Grp 2 9.7 (9.3-10.4 sleep onset times, shorter (o.30); Grp 3 11.8 (9.9-12.7 sleep periods, and shorter (0.45) / normal sleep times than younger subjects. Girls had longer sleep times and more motionless sleep than boys; increased family stress correlated with poorer sleep quality, increased daytime sleepiness assoc. with greater age and shorter sleep period per ACT

Sadeh (64) Observational study / out –of- 262 (220) M-115, F-105. Grp None / AMI AMA -32 / Rt ACT and Others, anthropo- NS / NS Newborns slept twice as lab hospital bed / entire 24 H, 1, 102, Grp 2, 118 / 20.4 and ankle / 1 min / NS / ASA metric measures much during nighttime hours 4C-b 1 D ACT anthropometric 20.4 / normal and other gesta- than daytime. Infants of ges- measures tional diabetes tational diabetic mothers (IGDM) had significant correlations between skinfold measurements and quiet sleep. Later gestational age

370 assoc. with increased quiet sleep percent

Sforza (39) Validation / laboratory / noc- 35 (35) / 54.8±1.6 (37-72) / PSG; Anterior tibialis EMG / PSG: PLM (via EMG); ACT: NS/Technical problems dur- ACT: EMG Movements – leg Conclusions differ from those turnal only for 7.5 h, 13 OSAS, 22 EDS or snoring Gaehwiler / upper Rt foot / 5 PLM ing recording/ Most pts Dx ACT not valid. R=0.78 but of Kazenwadel, et al., 1995 2B PSG/ACT(n= 10; CPAP titra- or RLS/PLMD sec / 0.1g / NS w/OSA ACT underestimates move- due to more sophisticated tion on night 2) 1 nt ments, especially those <3 sec analysis PSG/ACT (n=25) and < 50 mv but was reliable across 2 nights in 8/10 subjects thus not valid for DX PLMD but valid for Tx evaluation.

Trenkwalder (72) Controlled clinical trial, DB, 32 (28) / Grp1 53±9 (37-73), PSG, log, Self report, log, PSG, ACT, SBJ ratings NS / Other sleep disorders, No diff. b/t idiopathic & ure- ACT systematically underesti- cross-over design / Grp2 49±11 (29-66) / RLS, Observation by physician in psychotropic meds, Rx abuse mic RLS, L-Dopa only effec- mated PLMs relative to PSG, 1A Laboratory, Home / Uremia charge / Moroport, Rimkus / Hx / Small Grp size tive during 1st 4 h of sleep ACT confirmed PSG finding Actigraphy Review Paper—Ancoli-Israel etal Nocturnal; 1Nt PSG/ACT + leg / NS / NS / NS decreased power of between- of diminished effectiveness 2Nt PSG BL, 1Nt PSG/ACT Grp comparisons after 4 h. + 2Nt PSG Tx (L-Dopa), 1Nt PSG/ACT + 2Nt PSG PL (Placebo) Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Wicklow (55) Observational study / home / Ns (NS) M, 7, F, 14/ 36 (19- Sleep log, diary / Cambridge Sleep diary, log / ACT, others; Exclusion; receiving tx for Mean ACT sleep latency < 3D ACT audio recording of 60; 10.8) / insomnia Actiwatch / wrist / 1 min / NS self-report of obtrusive sleep disorder, medical prob- mean diary sleep lat, p < .001. 4 C-a intrusive thoughts at bedtime, / NS thoughts, pre-sleep arousal lems affecting sleep, psy- Mean ACT SE > mean diary sleep diary scale chopathology, no sleep onset SE, p < .008. Mean ACT problems/ subjects recruited TST > mean diary TST, from general population, not p<.004. ACT/ diary sleep presenting patients latency, r=0.419, p<.001. ACT/diary SE r = 0.194 (ns). ACT /diary TST r = 0.526, p <.001. ACT sleep latency correlated with rehearsal/planning thoughts and thoughts about autonomic functions. No correlation between diary sleep lat and any thought category.

Wilson (52) Validation, ACT compared w/ 40 (40) / 44.9±7.9 / Insomnia, Self-report, Questionnaire, log, ACT NS / Fibromyalgia, other ACT & log both showed low other techniques / Home / Chronic pain log / MML / ndom wrist / 15 medical problems / None SE, long WASO, low TST; 5D-b Nocturnal; 2D ACT/log sec / NS / AMI ‘94 No diff on ACT variables b/t high severity & low severity pain Grps 371 Actigraphy Review Paper—Ancoli-Israel etal Very high r (R squared = Very 0.939) between time zone shift and activity acrophase not compared shift; however, to any circadian reference standard. ACT appeared to detect TST expected reduction of induced by forced desynchrony protocol (poorer sleep when out of phase w/ temp ACTrhythm), suggests can detect a disturbance induced but by circadian pacemaker, no reference standard. ACT to PSG r SOn>0.80 in of 14 cases and r 11 in 10 of 14. SOff>0.80 Although main study design not blind, comparisons of ACT to DLMO and PSG ACT were blind. 71 % of w/in 18 min of sleep offsets PSG sleep offsets. ACT data scorable on 9 of 10 ACTNts. cor- SOn and SOff related 0.39 to 0.51 w/DLMO NS on on school Nt, but r’s weekend. Comments from Reviewer from Comments or D alertness with Tx. or D alertness with Morning bright light delayed ACT increased mesor peak, and strengthened rhythmicity Activity phase shifted in same direction, but often not same amount, as time zone change. Lower activity mean and dur at destination. Movement freq. (weakly) predicts SBJ SE, SQ; Circadian time of going to sleep affect- ACTed TST & SBJ SE & quality (all reduced when starting sleep between 1000h and 1300h), but did not affect movement index or freq. DLMO Phase of SOn, SOff, less dispersed after fixed light-dark than self-selected ACTlight-dark. SOn to SOff ACTr=0.72, SOn to DLMO ACTr=0.82, to DLMO SOff r=0.76 during self-selected but not fixed light-dark. Mel significantly correlated offset with age (r=0.62) and tanner stage (r=0.62). but not SOn, DLMO SOff later in 10th, more sleepiness on sleep onset REM in 10th paper Nt SDG slept more during SDG and D than MMNDG; had lower activity mesor, and more blunted amplitude, were more phase-delayed than MMNDG NS / NS / NS No improvements in Nt sleep NS/ NS / convenience sample, low sample size, no control Grp no sleep disorders/ Healthy, NS / low num of subjects, all not blind to time of D F, NS / irregular sleep, sleep disorders, illness, psychiatric disorder / NS psychopathology / NS NS / NS / Subject selection criteria NS ACT: D sleep, Nt sleep, ACT D sleep, Nt sleep, ACT: circadian acrophase, mesor, goodness of fit Phase shift of daily ACT: and activity onset, offset acrophase. Magnitude of motions/5-mins and activity dur (“alpha”) SOff, WASO, SOL, Log: BT, SOn, SOff, ACT: SBJ SQ; movement index, freq. TST, of movement onsets Log; mel; endocrine measures cortisol; PSG: SOn, SOff; MSLT SOn, SOff; ACT: TST SOn, SOff, ACT: NS / sleep disorders, illness, Outcome Measures Measures Outcome / Bias Inclusion / Exclusion Conclusion from Study total mins sleep, % ACT: % sleep, total mins awake, awake, num awakenings, length of each awakening, acrophase, amplitude, mesor, circadian quotient; Mood: Geriatric Depression Scale; Others: MMSE .003g / Cole, Kripke ≥ None / Actillume / wrist / 1 None / Action3 sleep 5 min / NS / parameter extended cosine none / Motionlogger / wrist / 5 mins / 5-mins, acrophase / daily activity onset, offset, motions/ where) / Gaewhiler / ndom wrist / 30 sec / NS / ACCORD software (UK) + Horne et al 1994 algorithm + custom algorithms ACT mel / Mini PSG, AMA- 32 (AMI) / ndom wrist / NS / NS / Sadeh et al ‘94 NS / NS / Sadeh et al ‘94 Comparison Measures / Measures Comparison apparatus / Actigraph / Epoch Length / Placement Sensitivity / Scoring Actillume / wrist / 1 None / min / et al 1992 118 (77) / 85.7±7.3 (60-100) / 118 Demented NH pts 6 (6) / 41.8 (27-56) / Normal, 17 records in 6 people, east- west vs. north-south vs. no travel 9 (9) / NS (19-20) / Normal log; (temp reported else- 19 (14) / M 12.7±1.0 (11.2- 14.1) F 13.1±0.7 (12.2-14.4) / normal 40 (26) / 15±0.5 / normal Log / Mini-ACT (AMI) / NS / Sample Size (Completed Sample Size / Age (Range) / Mean Study) Medical Conditions (60- NS (77) / Grp1 85±7.9 / 100) Grp2 87±5.6 (74-96) NH residents RCT, unblinded, parallel / NH unblinded, RCT, 10 ACT, / Entire 24 hs; 18 D Tx w/either morning bright D light, evening bright light, evening dim red light, daytime sleep restriction Observational study / transmeridian travel / Entire 24-hr, ACT17-32 D including Ds before, during and after travel ACT Compared w/ other techniques; Forced desynchrony / Laboratory (dormito- ry used as lab) / Entire 24-hs; BL 8Nt sleep midnight-0800; Sleep Dep.1: 26hs; Forced desynchrony: 17 cycles (27hr W D) sleep 9 hr, 18 hr; Sleep Dep.2: 26hr; Recovery sleep: ACT 9hr; continuous Nonrandomized, Controlled unblinded, clinical Trial, cross-over design / Laboratory and Home / Entire ACT 24 hs; continuous 7D home, throughout study, self selected schedule (salivary DLMO last night), 8D home, fixed light-dark schedule (salivary DLMO last night), 3D long (14 h) Nt ~36 hs modified con- w/PSG, stant routine w/ MSLT and PSG Observational study / laboratory and home / Entire 24 hs; ACT14 D and log and phone 1 message machine at SOff, MSLT, salivary DLMO, PSG, repeat in 9th and 10th grade w/ 1 h earlier school start time in 10th. Study Criteria Design / Criteria Design Study Protocol Location / Unblinded, nonrandomized, NH / observational study / ACT Entire 24-hs; 3D Evidence levels for studies involving circadian rhythms circadian involving for studies levels Evidence Ancoli-Israel (110) 4C-b Binkley (92) 4C-b Blagrove (87) 3C Carskadon (118) 2B Carskadon (119) 3C Table 4— Table AuthorCitation - / Level Evidence Ancoli-Israel (112) 4C-b

SLEEP, Vol. 26, No.3, 2003 372 Actigraphy Review Paper—Ancoli-Israel et al Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Cole (44) RCT, single blinded / home / 78 (59 assigned to Tx) / 25 Log, mel,/ Actillume / wrist / mel 6-SMT; ACT 18-40 yrs, ICSD criteria for Bright light and behavioral Tx ACT used to determine BL Entire 24 hs; 7D ACT BL, 5D (14-34) / DSPS-SOn before NS / NS / lab developed DSPS / non DSPS sleep dis- advanced the 6-SMT sleep phase and post-Tx sleep 4C-b ACT Tx (last 5D of 26 D Tx or after 0200hs. order, serious health problem, acrophase vs. dim light place- phase. Tx: bright/dim light period), 2D ACT f/u meds, alcohol abuse, Rx, seri- bo Tx, but post-Tx phases (26 D) and behavioral Tx ous psychiatric problems, were listed earlier. However, (systematic advance of bed- recent shift work or jet lag / pts with late 6-SMT acrophas- time and time of arising, NS es did have phase advances avoid daytime naps, minimize (vs. placebo) of 6-SMT and afternoon/evening light). SOn as well as decreased AM Light mask: 2700 lux / 57 lux sleepiness. when eyelid closed.

Dagan (57) Case series / Home / Entire 61 (NS) / 30.17±11.26 (16- None / MML / wrist / NS / Survey No meds , DSPS Dx, mel Tx, 96.7 % report mel helped No validation of ACT other 24-hs; 4-7D ACT BL to help 54) / DSPS NS / NS Sadeh; Other: zero- finished ≥ 1 yr prior to study DSPS, 91.5% relapse after than pt. report ACT used only 5D-a diagnose DSPS; f/u question- crossing mode / No control Grp mel cessation to aid initial Dx of DSPS. naire ≥ 1 yr after 6-wk mel Outcome assessed only by Tx questionnaire, not ACT.

Daurat (97) Controlled clinical trial; DB / 36 (24) / 51.2±2.2 / Normal Temp (n=19) / Actiwatch / Temp; ACT: sleep dur, activi- Healthy, no meds x3 mos, no Zopiclone improves ACT ACT detected both circadian Hotel rooms set up as lab / ndom wrist / 1 min / NS / ty index, ACT acrophase & time zone travel / Core temp identified sleep after trans- shift and sleep disturbance 4C-b Entire 24-hs; 6D ACT BL, 1D Actisom (P.Denise, France)- amplitude; Mood: VAS not unmasked meridian travel F(1,22)=6.3, temp BL, flight, 6D ACT post sleep visually scored, cosinor “mood”; Jet lag Sx VAS p<.05; The greater the dysyn- flight, 1+1D temp post flight, for phase “tired”, “ill-being”, “diges- chronization of CRs post- Zopiclone or Placebo before tive”, “energy” flight the shorter the sleep bed Nts 1-4 post flight dur. Zopiclone did not improve SBJ jet lag

Dawson (93) RCT, parallel design (blinding 36 (36) / 23.6±3.9 (18-30) / None / Gaehwiler / ndom Temp; mel; DLMO, ACT: NS / NS / NS Bright light shifted DLMO, ACT variables showed signif- not mentioned) / Lab / Entire Normal wrist / 30 sec / NS / Root movement index, total energy, reduced core temp during D icant differences between

373 4C-b 24-hs apparently, 2Nt DLMO mean square (RMS) activity energy during arousal, energy sleep, improved ACT SQ bet- light and placebo and often BL (Nt sleep) & core temp, times dur during non-arousal; ter than mel Tx or placebo. between light and mel Tx dur- 3D ACT (Nt work, D sleep) Behavioral: cognitive per- Mel didn’t shift DLMO, but ing D sleep in simulated shift & core temp, 24hs DLMO & formance improved sleep & reduced work. core temp (sleep not core temp somewhat. described) Random assign. to bright light, dim light, mel or placebo during shift work Ds

Dijk (32) Controlled clinical trial, DB, 5 (5) / (37-46) / normal PSG, log, mel urine, temp Log; ACT: SP time, TST; All astronauts / NS / Cannot Space flight associated w/ ACT detected apparent bias in nonrandomized / Space shut- core and ingestible sensor, neurobehavioral Assessment distinguish effects of space reduced ACT SP time, SBJ PSG sleep measurement 3C tle, pre, post flight / Entire 24 urinary cortisol / MML and Battery flight from those of work SQ, performance and delayed hs; ACT: 1-2 D 2 mos pre Actillumes / ndom wrist / NS schedule; small sample cortisol rhythm relative to flight; 1-2 D 1 mo pre flight; / NS / Cole/Kripke scheduled sleep. ACT Sdur 0-7 D 1 wk pre flight; 10-16 longer on PSG recording Nt; D in flight; 4 D PSG in flight; might reflect greater adher- Actigraphy Review Paper—Ancoli-Israel etal 5 D ACT post flight; 3 D PSG ence to sleep schedule on post flight; 0-3 D other meas- PSG Nt. No benefit of mel on ures post flight sleep observed.

Eissa (131) Observational study, Case 46 (46) / 42.9±21.8 / BP / MML / ndom wrist / NS ACT: W vs. sleep dichotomy Consecutive hypertension pts. Estimates of circadian or Actigraph-identified sleep series / Home / Entire 24-hs; Hypertension / NS / Cole / Possible daytime removal sleep-induced BP reduction assumed to be “true” sleep by 5D-a 1D ACT 1D ambulatory BP artifact could have overesti- (“dipping”) differ when “actu- authors monitoring mated D sleep (naps) al sleep” is defined by ACT than when it is defined by fixed time-of-D criteria Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Evans (88) Observational study / Home / 14 (14) / 81.2±5.9 (71-91) / log / Motion logger / ndom log; ACT: SOn, SOff, TST, Healthy elderly / sleep meds, Elderly have frequent Results could mean that ACT Entire 24-hs; 2D ACT & log Normal wrist / NS / NS / (AMI) soft- num naps, nap mins, num tranquilizers, sleep disorders, arousals, reduced SEI, short is valuable for detecting naps 4C-a ware arousals neuro or psychiatric problems sleep, naps more frequent not reported in log, but possi- / Artifact rejection not men- than expected; naps underre- ble that wakeful inactivity tioned so possible overestima- ported; satisfied w/ sleep was misscored as sleep by tion of D sleep; subjects ACT; findings of fragmented selected for no sleep com- nocturnal sleep and greatest plaints, so finding of satisfac- nap frequency in afternoon tion w/ sleep to be expected consistent w/ other studies’ findings w/ PSG in elderly

Glod (107) Case-control study / Home / 26 (26) / Grp1 11.0±3.3, Grp2 NS / Motion logger / Belt / 1 ACT: cosinor 24h & 12h har- SAD Dx, med-free, healthy / Children w/ SAD have less ACT rhythm distinguished Entire 24-hs; 3D ACT, com- 11.6±3.7 / Grp1 SAD, Grp2 or 5 mins / NS / NS monic and periodogram NS / NS robust circadian activity children w/ a psychiatric dis- 4C-b pare ACT rhythms in SAD vs. Normal rhythms & lower circadian turbance from controls; normal children amplitude than normal con- Dependent variable was activ- trols, but no phase delay of ity level (belt-worn monitor) activity not sleep

Gruber (113) Case-control study / Home / 102 (NS) / Grp1 9.6±2.7 (6- Log / NS / NS / NS / NS / Log; ACT: SOn, SP, SE, TST, DSM-IV ADHD / NS / NS More instability of ACT SOn, 5N ACT 14), Grp2 9.9±1.7 (7.5-11.5) / Sadeh longest sleep, quiet sleep (no SP duration, and TST in 4C-b ADHD vs. control motion), Nt awakenings ADHD than control. No difference in mean values. Only ACT, not log, detected higher standard deviations.

Guilleminault (121) NS / Home / 24-72 hs oral 12 (NS) / newborns / normal PSG / Vitalog / wrist / 2 mins ACT: longest inactivity NS / NS / some lost data. Rectal temp rhythm estab- temp at birth every 3 hs, At / NS / visually scored Rectal temp endogenous lished by 6 week age in 2 of 3C 3,6,8,16, and 20 weeks: 60 hs rhythm masked by rest-activi- 12 infants and by 10 week in to 7D ACT and rectal temper- ty rhythm. 12 of 12. Longest ACT inac- 374 ature tivity ‘closely related’ to longest PSG SP. Lowest temp occurred at time of longest ACT inactivity. Consolidated W only occurred after rectal temp rhythm established.

Heikkila (120) Case-control study / 24 / Grp1 24 (16-32), Grp2 7 Mel, temp, cortisol / (AMI) / Mel; Endocrine-cortisol; NS / NS / Possible masking Sleep-wake rhythm grossly No statistics on sleep Laboratory / 5D ACT and log, (3-10), Grp3 12 Grp4 NS / wrist / 1 min / NS / Cole ACT: SOn, % sleep of mel by light and temp by disturbed, but mel, cortisol 5D-a 24 hs serum mel and cortisol Neuronal Ceroid-Lipo-fusci- Kripke activity and temp rhythms usually and axillary temp nosis (NCL), juvenile NCL, normal in NCL patients. infantile NCL, Jansky- Bielschawski disease, Normal control

Actigraphy Review Paper—Ancoli-Israel etal Honma (184) Observational study / Mental 13 (8) / 84.9±7.8 (74-96) / NS / model not specified ACT: diurnal pattern, dom NS / NS / Small sample, high Identified 4 circadian “types” Actigraph identified 24h Hospital / Entire 24-hs; ≥ 10D Dementia & Delirium (AMI) / ndom wrist / 1 min / period dropout rate, SBJ analysis of pts, activity patterns varied rhythm in 8 of 8 cases, 5D-a ACT NS / visual & chi square peri- (except periodogram) w/ clinical manifestations of despite severe behavioral dis- odogram delirium; dom period near turbance 24h in all cases, (some had secondary 12h peak)

Huang (116) Observational study / Home / 65 (65) / young 24±4, middle- None / Actiwatch-L-plus Sleep diary, mean illumi- Healthy / physical, psychiatric Compared to young and mid- Entire 24-h; 5-7 D ACT aged 42±3, old 68±6, oldest (Cambridge Neurotechnology, nance, ACT TST, SE, SOL, or neurological illness, cogni- dle-aged volunteers, old and 4C-b 83±4 / none Cambridge, UK/ wrist / 1 min wake num, num naps in D, tive disorders, insomnia, alco- oldest volunteers had lower / NS / Rhythmwatch and fragmentation index, interdai- hol or drug abuse, extreme TST, SE, and circadian ampli- Sleep Analysis 98 Software ly stability (IS), intradaily morning or evening type tude, longer SOL, and more (Cambridge variability (IV), least-active 5 awakenings, sleep fragmenta- Neurotechnology) h (L5), nonparametric circadi- tion, IV and naps. IS did not an amplitude differ between the four groups, suggesting similar synchronization to Zeitgebers. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Jean-Louis (122) Unblinded, nonrandomized, NS (32) / 44.76±20.64 / SSS, log / Gaehwiler ACT: TST, SE index, WASO, NS / NS / Small sample size Women had a better sleep Relatively small sample observational study / Home / Healthy normal volunteers Electronics (Hombrechtikon, SOL, frequency of transitions profile than men by ACT 4C-b Entire 24-hs; 5D ACT Switzerland) / dom wrist / 60 between sleep and wakeful- sec / ≥.1g / ADAS ness, daytime activity level, auto- r (amplitude of activity) Jockovich (105) Controlled clinical trial, DB, 19 (19) / NS / normal None / NS / NS / NS / NS / ACT; SE, TST, SOL: emergency medical residents No effects of Mel Tx on D cross-over design /home NS Behavioral variables; SSS working night shifts / heavy sleep or Nt sleepiness or 5D-b /diurnal only for sleep period. (Stanford Sleepiness Scale) : alcohol or caffeine, opioids, mood 375 1 mg mel or placebo 30-60 Mood ; POMS (profile of benzodiazepines / mel tx tim- min before day sleep after Mood States) ing not optimized for circadi- night work. Two series of = 3 an effect night shifts separated by = 1 week. ACT during day sleep

Kario (185) NS / Home / Entire 24 hs NS / 48±8.6 (33-66) / mild, None / (AMI) / waist while ACT: weighted average activ- NS / cardiovascular events or Physical activity (i.e. ACT untreated hypertension awake, wrist while asleep / ity 6 mins before each BP used anti-hypertensive meds / motor activity is one of the 5D-b NS / NS / NS measurement (every 15 mins NS determinants of ambulatory during W, every 60 mins dur- BP and its diurnal variations. ing sleep); Ambulatory BP No dippers exhibited greater sleep activity than extreme dippers.

Kubota (98) NS / Home; Laboratory / Dim 6 (NS) / 26 (21-35) / Normal NS / (AMI) / NS / NS / NS / Temp NS / NS / NS Evening bright light exposure

Actigraphy Review Paper—Ancoli-Israel etal light (150 lux) vs. bright light NS for 2 ½ hs delays temp nadir 5D-b (3000 lux) from 19 – 21.30 from 04:08 to 05:29 and dis- each evening for 5 days. turbed subjective sleep.

Leary (129) Case series / Clinic / 24 h 434 (434) / 47.9 / hyperten- None / Gaehwiler / dom wrist ACT; Nocturnal fall (‘dip’) in Consecutive pts referred for Daytime motor activity ACT used to quantify D & Nt ACT + ambulatory BP moni- sion / 10s / 0.1 g / NS systolic and diastolic BP evaluations of hypertension / (ACT) was positively corre- activity levels. ACT was not 4C-b toring NS / NS lated w/ nocturnal dip in BP. an object of study. Higher Nt motor activity was negatively correlated w/ the nocturnal dip.

Lemke (134) Observational study / out-of- 16 (16) / 54 (24-65) / major NS / Acotmeter, Zak, ACT: mean activity level 7- 9 NS/ Parkinson’s & other Depressed in-pts displayed lab / 72 hr depression episode w/ melan- Germany / ndom wrist / 2 AM vs. 7-8 PM; Mood: mul- movement disorders / NS significantly greater motor 4C-b cholic features mins / 0.1 g / NS tiple affective adjective activity in the morning com- checklist pared to evening, but subjects felt more active in PM, and more tired & depressed in the AM. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Lockley (29) ACT compared w/ other tech- 49 (49) / 46.6±12.2 / Blind log; mel; Urine samples / log: SOL, SOn, SOff, Num Meds influencing sleep or ACT and logs yielded similar Unique use: to see if sleep niques / Home / Entire 24-hs; Motion loggers or MML / awakenings, dur awakenings, mel / NS / NS results for some aspects of (daytime nap num & dur & 3C mean nights/subject=23±7 wrist / 1m / NS / Action 3; num naps, dur naps; ACT: sleep but not others. night sleep dur) greatly (range 6-35) of ACT + log Other: Zero crossing mode SOL, SOn, SOff, Num awak- Proportion of pts showing dif- altered by CR type (normal enings, dur awakenings, num ferences between ACT and entrained, abnormal naps, total nap dur log greater than those show- entrained, or free run) ing no differences.

Lowden (186) Controlled clinical trial, 23 (19) / 42 (31-59) M, 15; NS / AMI / non-dom wrist / 1 Sleep diary/log, Act: BT, NS /NS Staying on home time greatly ACT detected numerous unblinded, randomized, F, 4 / NS / NS / Action 1.24 wake-up time, time awake, reduced jet lag symptoms and effects of flight pattern and 4C-b crossover design / NS / entire naps, TST; Behavioral; sleepiness during layover but timing of predicted sleep- 24 h. Two trips each with 1 Karolinska Sleepiness Scale; not after returning home wake and detected treatment tx in counterbalanced order. Others: jet lag rating effect on time spent awake Tx 1: maintain home sleep before main sleep time. Tx 2: adopt local sleep time. 10 D ACT (3 D BL), 1 D air travel 9 h east, 2 D layover, 4 D f/u

Luboshitzky (187) Case-control / Home / Entire 46 (NS) / Grp1 72.7±6.1, Mel / Mini-ACT AMA-32 ACT: SOn, SOff, SP time, NS / NS / 6SMT pooled into 6SMT lower in AD and nor- Study not designed to meas- 24 hs; 48 h urine 6SMT, 5 D Grp2 61.0±5.7, Grp3 (AMI) / wrist / NS / NS / TST, SE, longest continuous all D and all Nt samples so mal elderly than young. ure 6SMT phase. ACT not 4C-b ACT, 5 D light measurement 33.3±8.4 / Alzheimer’s dis- ACT Scoring Analysis, other: sleep, WASO, min inactivity; low resolution Compared to elderly and adequately compared to refer- ease, healthy elderly, healthy light meter on shirt Behavioral: MMSE; light young normals, AD pts have ence standard. young exposure poorer sleep. No relation between 6SMT and sleep. SOn and SOff earlier in AD and elderly normal than young normal. 376 Luna (103) Observational study / 14 (9) / NS / Normal shift log / ‘model #32’, (AMI) / Log: TST Rapid rotation shift work (air Masked temp retained diurnal ACT may have identified Workplace / Entire 24-hs; workers wrist / NS / NS / General temp: visual inspection of traffic control), Present for orientation on Nt shift. unreported on-duty naps in Nt 4C-a 21D ACT, oral temp every 4h, activity analysis program plot, ACT: total activity, dur of study / NS / 27% of Subjects felt more fatigue & shift air traffic controllers. sleep loss, mood and perform- (Elsmore) Behavioral: collected insuffi- ACT data lost due to techni- confusion and less vigor on ACT & self-report agreed ance rated daily at mid-shift cient data, Mood: POMS cal difficulties; temp masked; Nt shift, activity mean lower there was more sleep during problems w/ data reduction on Nt shift, More ACT identi- Nt work than on other shifts. for home sleep analysis fied sleep at work on Nt than D sleep after Nt shift was not on D or swing shift worse than sleep after other shifts, possibly due to scoring methodology.

Lushington (188) Longitudinal study / NS / 2 NS (NS) / 23.5 (18-31) / nor- NS / NS / wrist / NS / NS / Mel urinary; log NS / NS / NS Urinary mel onset from a sin- Log used to determine SOn wks log & ACT mal Gaehiler Electronics gle Nt can be used to predict times & these were then 4C-b subsequent onset times w/ in “confirmed” from ACT times ± 97 mins. A close temporal Actigraphy Review Paper—Ancoli-Israel etal relationship was also formed between mel onset and SOn

Mansoor (128) NS / NS / NS NS (NS) / 57 (10) / untreated None/ MML (AMI) / dom ACT: mean & peak ACT NS / NS / NS Correlation of motor activity, hypertension wrist / 1 min / NS / NS before each BP/HR reading BP and HR is highly variable 4C-b (every 5 mins); Ambulatory from pt to pt. Nondippers BP, HR have higher sleep activity than dippers.

Martin (109) NS / NH / Agitated Behavior NS (74) / 82.5±7.6 (61-99) / NS /Actillumes (AMI) / NS / ACT: max + mean activity; NS / stroke, psychiatric dis- The mean acrophase for agi- Rating Scale (ABRS) x 60.5 Alzheimer’s disease 1 min / NS / Action 3 (AMI) Agitated Behavior Rating order, pre-dating dementia / tation was 14:38. Only 2 pts 5D-b h; ACT x 3D (72 h) Scale; and Illumination Level NS (2.4%) were “sundowners” Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Martin (132) Observational study / NS / NS (NS) / 58.3±9.8 (45-76) / Log / Actillume (AMI) / non- ACT: Sleep, wake; MMSE, NS/ NS / NS There were important distur- ACT x 3D; entire 24-hs Schizophrenia dom wrist / NS / NS / Action QWB, Brief Psychiatric bances of CRs and S/W, 4C-b 3 (AMI) Rating Scale, SAPS, SANS, which were related, cognitive AIMS, EPS, neuropsycholog- functioning and low levels of ical examinations, light expo- illumination exposure sure

Middelkoop (41) Validation / Home / 45 con- 20 (20) / 24.1±3.7 / Normal NS / GAHWILER electronic ACT: Activity level NS / NS / Healthy young Wrist placements detected Wrist placement superior to secutive hs starting at 12:00; CH-8634 Hombrechtikon / Lt adults only more activity counts than ankle or trunk. Dom wrist 4C-b 45 hs ACT+log wrist, Rt wrist, Lt ankle, Rt ankle or trunk placements. appears better for discriminat- ankle, trunk (navel) / 5s / NS dom wrist placement yielded ing level of waking activity. / NS greater diurnal activity counts than ndom. All sites clearly showed circadian sleep-W differences. Suggests more study w/ placements concomitant w/ PSG

Middleton (89) RCT, DB, crossover design / 10 (8) / 23.9±0.75 / Normal Mel sulphatoxy, core temp log; core temp-(demasked); Normal mel level & timing / Mel Tx phase-shifts sleep & Circadian phase of wrist Lab / 2x30D free-run in 4 lux; (demasked) / MML (AMI) / mel sulphatoxy acrophase and NS / Small sample, but prob- core temp rhythms. Mel Tx activity and demasked core 2B mel Tx first 15 Ds at 20:00, wrist / 20 sec / NS / Cosinor period; ACT: acrophase, SOn, lem offset by good experi- usually synchronized the S/W temp generally agreed, pro- crossover to placebo Tx for & spectral analysis; Action3 SOff mental design rhythms but only inconsis- viding OBJ evidence that 2nd 15 Ds. Separate 30 D w/ sleep analysis Cole tently synchronized the core ACT can provides useful cir- placebo 1st 15 days. temp rhythm cadian phase measurements Continuous ACT, rectal temp, urinary mel sulphatoxy, logs

Middleton (90) Observational study / Lab / 6 (6) / 26±2.7 / Normal log; mel sulphatoxy; core log; temp; mel sulphatoxy; NS / NS / Temp not Mel, temp & activity rhythms Activity rhythm usually Entire 24-hs; 21D free-run in temp / MML (AMI) / wrist / ACT: SOn, SOff, TIB, TST, demasked, small sample free-ran in 5 of 6 individuals. agreed w/ the mel & temp 377 2B L/L 4 lux & knowledge of 20-30 sec / NS / Cole, regres- WASO, SEI, activity In the 6th, only mel free-ran. rhythms. Agreement w/ the clock time continuous ACT, sion on cosinor acrophases, acrophase; Behavioral: per- Temp may have remained at mel rhythm provides OBJ rectal temp, urine every 4h, spectral analysis formance 24h period due to masking by evidence that ACT can be a performance every 3h while the rest-activity cycle. useful phase marker. awake Agreement w/ temp provides weaker evidence because it may be due to masking.

Mishima (99) RCT, unblinded, cross-over 22 (NS) / Grp1 81, Grp2 78 None / (AMI) / ndom wrist / ACT: Total, daytime & night- DSM IV Dx of vascular Vascular Dementia: daily Shows ACT detected Tx design / NH / Entire 24-hs; / Grp1 Vascular dementia, 1 min / NS / activity counts time activity, % night/total dementia or Alzheimer’s / bright light reduces nighttime effect in well-designed, con- 4C-b continuous ACT for 1 wk pre- Grp2 Alzheimer’s dementia used, sleep not scored activity mixed dementia excluded / activity; Alzheimer’s trolled trial. Suggest ACT Tx, 2 wks Tx, 1 wk post-Tx, Unable to blind Tx, small N Dementia: bright light does useful to detect outcome in ≥ 4 wks washout cross-over; not affect activity rhythm CR Tx study Tx=5-8000 lux 9-11am; Control=300 lux 9-11am Actigraphy Review Paper—Ancoli-Israel etal Mishima (125) NS / NS / Entire 24 hs; 7 + 41 (NS) / Grp1 6.4±7.6, Grp2 Temp ambulatory Rectal / Temp; ACT: mean total daily NS / mixed (SDAT and MID) The SDAT grp showed posi- 24 hr Ds. SDAT & MID grps 79.1±5.6 / SDAT (N=20), MML (AMI) / nondom Wrist activity, mean D activity, dementia, severe motor dis- tive correlation between 4C-b compared MID (N=21) / 1 min / NS / NS mean Nt activity, % Nt to D turbances, sleep disorders /NS activity (total daily activity activity and %Nt activity) and dementia severity. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Mormont (150) Unblinded, nonrandomized, 200 (192) / 58 (20-75) / None / ACT (AMI) / NS / 1 Endocrine measures: circadi- NS / Poor general health Pts w/ marked rest/activity observational study, cohort Ambulatory metastatic col- min / NS / NS an changes in cortisol & (World health Organization rhythms had a 5-fold higher 4C-b study, longitudinal study / orectal cancer pts referred for WBC; ACT: auto-r at 24h & a performance status 72) / NS survival rate at 2 yrs, had a 5- Home / Entire 24-hs; 3D ACT chronomodulated chemother- dichotomy index comparing fold higher survival rate at 2 apy amounts of activity when in yrs, had better QOL & report- bed & out of bed, mean activ- ed less fatigue. Rest/activity ity; Mood: HADS (scale of rhythm remained a significant anxiety and depression); predictor of survival in multi- QLQ-C30 - Quality of Life variate analysis. Scale for European Organization for Research & Treatment of Cancer

Nagtegaal (60) Case report / home / entire 24 1 (1) / 15 (NS) / head injury Self report, mel, temp; appa- Temp, mel, ACT NS / NS /NS Etiology of DSPS was pre- ACT was used to “quantify h; 24 h ACT before Tx w/ ratus: Gaewiler Electronic; sumed to be head injury. her S/W behavior,” but no 3C mel, after Tx w/ melatonin at nondom wrist; NS DSPS responded favorably to quantitative analysis was 03 30 & after Tx at midnight mel done. By visual inspection ACT period of low activity corresponded to period of mel secretion and low rectal temp

Otsuka (178) Validation / Laboratory; Out- 44 (44) / Grp1 (28-46), Grp2 ECG; BP; HR / Activetracer / ACT: Activity level; BP Simultaneous ACT & BP can Hard to pull data out of this of-lab hospital bed / Entire (28-78) / Normal; Healthy waist / 1 sec / 0.01 to 0.50g / be used to show relationship article; uses a lot of individ- 5D-a 24-hs; 72+hr ACT 7 ECG, NS of S/W w/ BP. Suggests S/W ual examples. (also BP for 48hs & HR for rhythms of about 7 Ds (cir- 24hs) caseptan) in addition to circadian & a 3.5 D (circasemisep- tan) rhythm in irregular S/W

378 cycles but able to identify individual differences in all these

Park (106) Observational study/ 12 (NS) M /32.4 ± 6.8 (23- None/AMA32 cl by AMI / ACT : TST, BT, wake-up, NS / Small sample TST on nt shift < D shift < Authors state that ACT results home/entire 24 h 44) / rotating 3 shift workers, non-dominant & wrist /1 min num nap, nap dur, amplitude, evening shift. TST decreases correspond to their previous 4C-b age < 25, 26-35, >36 / NS /Cole/Kripke and other activity on duty with age PSG results in nt workers was zero crossing mode

Pollak (123) Case-control study / NS / NS (86) / Grp1 80.7±7.9, NS / MML (AMI) / ndom NS Age ≥ 65, occurrence of dis- Demented older people were Entire 24 hs; 9D ACT Grp2 73.7±7.2, Grp3 wrist / NS / NS / NS ruptive nocturnal behaviors / significantly less active in the 4C-b 7.3±10.2, Grp4 58.1±16.0 / NS / NS daytime than their caregivers, NS and circadian rest–activity amplitudes were smaller

Poyurovsky: (133) Case-control study / NS / NS (NS) / Grp1 36.4 (19-50), None / ACT AMA 32 (AMI) / ACT: overall motor activity, NS / Psychotic agitation / NS Schizophrenic in-pts w/ neu- Actigraphy Review Paper—Ancoli-Israel etal Entire 24 hs; ACT x 24 hs Grp2 29.6 (18-49) / NS / 1 min / NS / 0.1 g / NS SL, sleep dur, sleep continu- roleptic induced akathisia 4C-b Schizophrenia w/ or w/o neu- ity, SE; Barnes akathisia scale were motorically more active roleptic induced akathisia from 11:30 to 14:15 and from 18:00 to 21:00 than those w/o neuroleptic akathisia.

Quera-Slava (94) Nonrandomized / workplace / NS (40) / Grp1 36±7, Grp2 NS / Gaehwiler / wrist / NS / log; mel (urinary 6-sulfatoxy NS / NS / NS Night-shift workers slept 5-7 D 35±7 / normal NS / NS mel); ACT more on Ds off. (ACT inac- 4C-b tivity). Some night-shift workers are able to switch acrophase of urinary mel (to 1208 hr, mean)

Quera-Salva (95) NS / laboratory (2Nt) / NS 40 (40) / Grp1 36±7, Grp2 log / actigraph/photometer log; mel (urinary sulfatoxy 25-55 yrs of age good health; 6/20 (30%) of night shift 35±7 / normal; 20 on fixed (Gaehwiler Electronic, –mel); ACT: TST; No Rxs affecting mel, not nurses were physiologically 4C-b night shift, 20 on fixed D Physiocom, Paris, France) / Performance (Stemberg anxious, depressed / NS / NS able to adapt to a fast-shifting shift ndom Wrist / 1 min / NS / recovery test for seven letters; S/W schedule. This shift was Actison (Axon. Physio, com. 4-choice reaction time test. associated w/ a change in the Paris, France) Urine samples (Q2h x 24 hr) acrophase of 6-sulfatoxy mel Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Reid (25) Validation; ACT compared 32 (32) / Grp1 21.2±2.7, Grp2 PSG / Gaehwiler / ndom hand PSG: Sleep, wake SOn, SOff; NS / Smokers, taking meds, ACT valid for S/W activity & w/ other techniques / 43.9±6.8 / NS / 30 sec / 0.1g 125 ms sam- ACT: Sleep, wake, SOn, SOff sleep disorders / Lab shift sleep dur (although ACT dif- 3C Laboratory / Diurnal; Entire pling time (8 kg) Baudpass work studies may not com- ferent from PSG for SOff) but 24-hs; ACT+PSG (when filter 0.25-3.0 kg / NS pletely reflect real world, not as good for more specific sleep) for 1Nt adaptation, 2D their older subjects were real- measures such as SE; ACT: of 12 hr D shift (night sleep), ly middle-aged. agrees more with PSG for 2D 12hr night shift (D sleep), young adult than for middle- 1 1/2D “off” aged because more quiet W scored as sleep in latter, as the likelihood of sleep decreases so does ACT accuracy.

Sakurai (108) Observational study, ACT vs. 35 (35) / (65-95) / Normal log, illumination / Actillume / ACT: activity mesor, ampli- Healthy elderly / NS / No activity mesor and amplitude Mesor results clear, but other Light / Home / Entire 24-hs; ndom wrist / NS / NS / Cole tude, continuity of activity; mention of exclusion of off- correlate negatively w/ age in results not explained in suffi- 5D-a 3-4D ACT & light & logs, et al. & Cosinor (Action3) illumination (lux) wrist artifact, all Ms 65-95 range, decreased “sleep cient detail to interpret. Phase nurse interview at home amplitude” in 80-95 range, & goodness of circadian fit about 1/2 of subjects who NS show “discontinuous movement” have low light exposure

Samel (189) NS / laboratory / Entire 24 hs; 4 (4) / (23-29) / normal None / piezoelectric Temp: level, circadian, amp. NS / NS / NS CO2 levels up to 1.2 % (as 29D ACT conditions: normal accelerometer / ndom wrist / Acrophase; ACT: sdur., activ- occur during manned space 4C-b CO2 concentration and ele- NS / 1.0 g / NS ity level; mood: fatigue ques- missions) do not impair circa- vated CO2 tionnaire; salivary samples dian temp or S/W playtimes urinary cortisol, cate- but do ↓ D mean activity cholamines, (6OH mel sul- level & D man temp and cir- phate) cadian amplitude 379 Satlin (91) ACT compared w/ other tech- Grp1 28 (28) Grp2 10 (10) / Rectal temp / (AMI) acti- ACT: mean and % D and Nt Probable Alzheimer’s / psy- Alzheimer’s temp rhythm PLMs could account for some niques / Hospital bed / Entire Grp1 71.4±5.4 (61-82), Grp2 graph, model NS / Grp1 activity, circadian variability, chiatric, neurological, acute delayed but normal amplitude findings because only pts had 3C 24-hr; 72 hr ACT and rectal 72.8±5.9 (67-80) / Grp1 ankle, Grp2 waist / 5 mins / acrophase, mesor, amplitude, medical disorders / different and 24-hr entrainment. actigraph on ankle. Both temp Alzheimer’s, Grp2 normal NS / mean D and Nt activity, deviation from 24-hr rhythm actigraph placement in pts vs. Alzheimer’s activity rhythm activity and core temp cosinor, variance spectra, controls, temp rhythm masked delayed, more variable, lower acrophases were delayed in interdaily stability, circadian by activity amplitude, more nocturnal pts, but not clear if source is correlation activity. circadian pacemaker or mask- In pts w/ large phase angle ing. between activity and temp, more sleep fragmentation and lower temp amplitude

Scher (86) NS / NS / Nocturnal 2Nt ACT NS (NS) / 1 yr±14D / normal None / MML (AMI) / L ankle ACT: SOT, Sdur, num of Normal, full-term infants / NS Results did not support a gen- / NS / NS / NS awakenings, activity level, / NS eral link between S/W regula- 5D-b SE; Temp rating tion in infancy and tempera- Actigraphy Review Paper—Ancoli-Israel etal ment characteristics

Shapiro (127) Observational study; longitu- NS (104) / 67.0 (55-79) / NS None / MML (AMI) / dom Ambulatory BP, HR, diary Healthy + active / neurologi- Activity was not correlated w/ dinal study / home / 2 D wrist / 1 min / NS / NS (time, posture, activity) zero- cal cardiovascular, renal, BP or HR between subjects, 5D-b ACT (48hs) crossing mode endocrine + psychiatric waking or sleeping. Within Disorders, & med affecting subjects, activity was weakly cardiovascular system / NS correlated w/ HR & BP

Shochat (161) Observational study, ACT vs. 77 (66) / 85.76±7.3 / Illumination, Dementia rating ACT: % sleep & W, num NS / NS / NS Low illumination in NH pts, Light exposure / NH / Entire Institutionalized elderly, scale / Actillume / wrist / 1 naps, mins W between maps, higher light in D associated 4C-b 24-hs; 3D ACT and illumina- dementia (all except 3 of the min w/ 10 sec sub-epochs for num night awakenings, dur of w/ fewer Nt awakenings, but tion monitoring completers) max activity (i.e. max 10 each awakening; median & not higher D activity, nor D sec/min) / NS / NS, but mean illumination, min > 1k wakefulness, severe dementia Ancoli-Israel et al. 1997 NH lux & > 2k lux predicted more D sleep, light algorithm implied acrophase correlates w/ activity acrophase, more mins/D bright light predicts later activity acrophase Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Siegmund (85) Observational study / home / 39 (39) / (4wks-62yrs) / nor- Observation, tape recordings, ACT: no formal variables NS/ NS / NS In infants circadian rest-activ- Entire 24 hs, 7 D ACT mal photographs, video / actome- described ity patterns develop out of 4C-b ter, ZAK, (Gmblt, Simbach, ultradian components. Adult Germany) / ndom Wrist / 2 rhythms well related to natu- mins / NS / AKTOGRAPH ral light-dark cycle. ZAK Gmblt, Simbach, Variability of sleep/rest termi- Germany nation smaller than its onset. Mean sl dur. Of infants aged < 11 mos 9-12 h/D; adults 7- 10 h/D. Women often slept longer than men

Van Someren (100) NS / out-of-lab hospital bed / 29 (22) / 79±2 (64-97) / None / self-made / NS / NS / ACT: “Interdaily stability”. SDG pts / NS / NS Brightened whole-D light No reference to standard Entire 24 hs; 5 D ACT bright Dementia (16 AD, 3 MID, 2 1.0 g / NS (IS) “Interdaily variability”, (436→1136 lux) ↑ IS (↑ cou- 4C-b ambient light (mean 1136 lux) Alcoholism, 1 normal pres- (IV) amplitude of rest-activity pling of rest-activity rhythms vs. BL (436 lux) sure hydrocephalus severe rhythm (AMP) to environmental zeitgebers), visual deficiencies in 5 pts ↓ IV (↓ fragmentation) more stable rest-activity rhythm but no change in AMP. These responses prevented by severe visual deficits. Van Someren (101) Longitudinal study / Home / NS (10) / 73±1.5 / normal None / self-made / wrist / NS ACT: Daily +hourly variabili- Had to be healthy / NS / NS Aerobic training reduces ACT data are inadequately Before aerobic training 5 ½ D / NS / NS ty of activity & amp. Of cir- “fragmentation” (h-to-h vari- analyzed. Paper was poorly 4C-b ACT; Aerobic Training for 3 cadian rest/activity rhythm ability) of the rest- activity referred. mos (10 Subjects/ no training rhythm in healthy elderly 8 controls); After training 5 ½ men. D ACT; 1 yr after training 5

380 ½ D ACT

Van Someren (190) RCT, parallel, SB / NH / Grp1 19 (14), Grp2 8 (8) / none / Apparatus: self-made; ACT: IS, IV, RA Include early stage probable Higher IV, lower RA, and No reference standard. Entire 24-hr, 4D ACT BL, 4D Grp1 84±1.5, Grp2 77±3.5 / wrist; NS; Scoring: interdaily Alzheimer’s / Exclude neu- trend for lower IS in 4C-b ACT after 6 wk Tx, 4D ACT Grp1 Alzheimer’s, Grp2 stability (IS), intradaily vari- roleptic meds, visual deficien- Alzheimer’s vs. normal elder- at 6 wk follow-up; Tx = Normal ability (IV), relative ampli- cies /low sample size ly. TENS increased IS in TENS or placebo tude (RA) Alzheimer’s, but did not improve IV or RA.

Van Someren (111) Open trial/ NH/ compared 4 and 17 (NS) / NS/demented None/ 1) Actillume; 2) home ACT: Compared cosinor, NS / NS Non-parametric variables, data analysis methods for 2 made/ 1) non-dom wrist; 2) complex demodulation peri- especially interdaily stability 4C-b studies: Study 1: 28D ACT non-dom wrist /NS / NS / odogram, autocorrelation, and 24-h autocorrelation are during 5D BL, 10D bright Sleep NS interdaily stability, intradaily more sensitive to circadian light tx, 13 D F/U/ entire 24 h variability, relative amplitude disturbance at rest-activity rhythm than parametric analyses such as cosinor Actigraphy Review Paper—Ancoli-Israel etal

White (130) DB; randomized; cross-over 85 (75) / 57.8±9.1 / Stage I or None / MML (AMI) / wrist / ACT: sleep vs. W; II hypertension, systolic BP > Morn. & even. Times of ACT was used to determine design / home / Entire 24 hs II hypertension NS / NS / Cole et. al. Ambulatory BP, HR 200, diastolic BP > 110, administration of nisoldipine sleep & W times by the built- 4C-b Bradycardia, Tachycardia, (a long-acting dihydropyri- in algorithm of the AMI Recent stroke or MI, renal or dine Ca++ channel blocker) recorder or by use of an event hepatic dis; uncontrolled dia- have similar effects on 24 hr marker built on activated by a betic / NS / NS BP and HR. However, com- pt when retiring or awakening parison of sleep & W periods revealed differential BP and HR effects for morn. Vs. even. dosing

Wirz-Justice (81) Observational study, Case 7 (NS) / 50.0±6.5 (38-57) / None / Gahwiler / non-dom Visual inspection of activity single-drug therapy for schiz- Four patients taking classical series / Hospital or Home / Schizophrenia wrist / 1 min / NS / peri- plot, “best tau,” (circadian ophrenia for at least 1 yr, neuroleptics had disturbed 5D-a Entire 24-hs; 3-7 wk ACT odogram period), “best omega” (circa- compliant with actigraphy / circadian rhythms, three tak- dian amplitude) NS / small sample size, Tx ing clozapine had highly reg- not randomly assigned, no ular rhythms placebo Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Youngstedt (43) RCT, unblinded parallel Grp1 72 (72), Grp2 30 (30) / Log, mel (6-SMT) / Actillume Log; PSG: SOL, TST, WASO, Young: healthy good sleepers, Elderly w/ insomnia or Directly compared ACT to design/ Home / Entire 24-h, Grp1 68.0±4.3 (60-79), Grp2 (AMI) / wrist / NS / NS / Jean AHI, myoclonus index; ACT: elderly: insomnia or depres- depression had poorer syn- mel reference standard for CR 1A Home: 5-7 D ACT, light mon- NS (20-40) / Grp1 elderly w/ Louis 2001 SOL, TST, WASO, BT, W sion / apnea, acute health chronization between mel and (evidence level 1A); indirect- itoring, and log, 2 x 24-h mel insomnia or depression, Grp2 times, mid-sleep time; problems, high use of meds sleep (both home ACT and ly compared ACT to PSG urine 6-SMT; lab: 30 H urine young normal Behavioral: Insomnia self-rat- that affected mel / Results for lab PSG), earlier self-selected (level 3C). Home ACT phase 6-smt, 5 D PSG under ran- ing, SBJ sleep; Mood: elderly w/ insomnia or sleep times (only measured at validated by agreement w/ domly assigned lighting con- Depression CES-D; Other: depression may not be same home w/ ACT, correlated w/ mel phase. Home ACT sleep ditions illumination acrophase and as for general elderly popula- earlier mel acrophase), lower results generally agreed w/ illumination mean 4 h before tion TST and higher WASO (both lab PSG results, except on BT, 4 h after arising home ACT and lab PSG) SOL. Circadian malsynchronization correlated w/ poorer sleep in lab. Yoon (102) RCT, unblinded, cross-over 12 (NS) / (21-24) / normal None / MML / left wrist / NS ACT: SP time, SL, TST, SE; Shift work nurses / NS / vol- Day sleep and alertness dur- ACT identified sleep showed design / workplace / Diurnal; shift workers / NS / Action3 Behavioral: Backward unteers not blind to differing Nt work improved most significant effect of circadian 4C-b 4Nt night work, Tx during Nt Masking Test, Digit symbol ences between 3 Tx by bright light plus a.m. sun- rhythm Tx; ACT results in D 1,2,3; $ D ACT during day substitution test; Alertness glasses, followed by bright sleep agreed w/ SBJ alertness sleep performance tests N 1,3, VAS light alone. No effect on per- result in Nt work. and 4; 4 Nt alertness ratings; formance. repeated for each of 3 treatments, 1 mo apart: 1. Room light, 2. Bright light, 3. Bright light and a.m. sunglasses 381 Actigraphy Review Paper—Ancoli-Israel etal In subsample of 10 subjects, validated ACT w/ observations: 92% agreement Rather than rest or sleep, called variable “arousal” No normal controls for comparison Did not look at standard measures of sleep Comments from Reviewer from Comments other NH generalize to Can’t populations ACT just at night; obser- used vations of sleep used during the D % of ↑ In subsample of 10 subjects, validated ACT w/ observa- At f/u, tions: 92% agreement. intervention Grp had sleep SDG slept more during Nt and D than MMNDG; SDG had lower activity mesor, more blunted amplitude, and were more phase-delayed than MMNDG Menopause pts had more arousals & greater sleep disruption fatigue negatively correlated w/ activity The num of nighttime awakenings had the strongest assoc. w/ cancer-related fatigue. Subjects w/ higher fatigue had lower amplitude & lower peak activity paper No improvement in sleep phys- associated w/ improved ical function Pts on meds had higher means for sleep but not significant Minutes awake: 555D, 396Nt; Minutes asleep: 130D, 325Nt; awake: 81%D, %Time 56%Nt; % time asleep: 19%D, 44%Nt; num of transitions from wake to sleep: 22D, 37Nt; Mean length of awakening (mins): 33D, 10Nt; Mean lux: 63D, 2Nt 60 / NS / ≥ Incontinence / Coma, severe All aggression stay < 3 mos / subjects incontinent; leads to question of generalizability NS / NS / Subject selection criteria NS NS / OSA, PLMS / Possible selection bias, unclear cause not well described NS / NS / NS Fatigue higher during chemo; I or II breast cancer, Stage scheduled to begin chemotherapy after breast English-speaking, surgery, Karnofsky score NS Drop-outs had most severe symptoms In physical restraints, inconti- In physical restraints, to nent / Coma, inability ter- respond, combativeness, w/ minal illness / Only pts incontinence or in physical restraints Incontinence / Comatose, unable to respond, combative / Only incontinent pts Nursing home residents / NS / low sample size, subject selection criteria NS ACT: % sleep, max dur of ACT: sleep episode, mean dur sleep episode sleep, % total mins ACT: sleep, total mins awake, % awake, num awakenings, length of each awakening, acrophase, amplitude, mesor, circadian quotient; Mood: Geriatric Depression Scale; Others: MMSE log; ACT: arousal/non-arousal; Mood: POMS, STAI amplitude, num mesor, ACT: wakes; Other: Piper Fatigue Scale amplitude, peak mesor, ACT: nighttime awaken- activity, ings; Revised Piper Fatigue Scale (to measure cancer- related fatigue) Outcome Measures Measures Outcome / Bias Inclusion / Exclusion Conclusion from Study % sleep, dur of TST, ACT: sleep sleep episodes, longest episodes dur % sleep, mean TST, ACT: sleep episodes, peak dur episodes ACT: mins awake, mins mins awake, ACT: asleep, % time awake, % time asleep, mean length of awakenings, mean lux exposure .003g / Cole, Kripke ≥ .003g / Ambulatory .003g / ≥ Observation / NS / Dom wrist / 2 mins / NS / NS 1 Actillume / wrist / None / min / et al 1992 NS Actillume / wrist / None / / Questionnaire; log / Gaehwiler / ndom wrist / 30 sec / NS / ref to older paper, but no details None / NS / ndom / NS / NS / NS Questionnaire - Fatigue / MML / ndom hand / 5 sec Action3 every min / NS / (AMI) Comparison Measures / Measures Comparison apparatus / Actigraph / Epoch Length / Placement Sensitivity / Scoring dom None / 1C Sensors / NS wrist / 2 mins / NS / None / NS / dom wrist / 2 mins / NS / NS Monitoring (Actillume) software 29 (NS) / Grp1 88.6±10.4, Grp2 88.3±5.7 / Incontinence NS (77) / Grp1 85±7.9 (60- 100) Grp2 87±5.6 (74-96) / NH residents 25 (25) / M 89.0±4.2, F 86.5± 6.2 / NH residents NS (28) / 46.8 (40-55) / Menopause 72 (60) / 49.5 (30-69) / Breast Cancer 72 (60) (30-47) / 49.5±8.64 (33-69) / Normal, Breast cancer pts Sample Size (Completed Sample Size / Age (Range) / Mean Study) Medical Conditions Grp2 65 (65) / Grp1 84.4±7.2, 85.1±7.6 / Incontinence 186 (176) / Grp1 86.2±7.1, Grp2 86.1±9.1 / Incontinence RCT; unblinded; parallel RCT; design / NH / Nocturnal only ACTfor 12 hs; 5Nt BL, 5D ACT Observations BL, 5Nt f/u, 5D Observation f/u; Intervention: physical activity during D and extra quite and dark at night w/ decreased interruptions to change dia- pers Unblinded, nonrandomized, observational study / NH / ACT Entire 24-hs; 3D Unblinded, nonrandomized, observational study / NH / ACT Entire 24-hs; 1D Unblinded; nonrandomized; observational study; case-control study / Home / Entire 24- hs; ACT 1 wk Unblinded; nonrandomized; observational study; prospective, descriptive repeated measure / Home / Entire 24- ACThs; ACT 96hs chemo 1, 72hs midpt chemo cycle, repeated for 3 cycles of chemo Unblinded, nonrandomized, cohort observational study, longitudinal study, study, repeated measures / Home / ACTEntire 24-hs; performed ACT6 times, 4D at beginning of chemo for 3 cycles, 3D ACT at midpt between chemo for 3 cycles Study Criteria Design / Criteria Design Study Protocol Location / unblinded, parallel RCT, only design / NH / Nocturnal ACTfor 13 hs; 2Nt BL, 2Nt ACT final wk intervention Observational study / NH / Nocturnal only for 13 hs; 2Nt ACT Evidence levels for other clinical studies clinical for other levels Evidence Alessi (164) 4C-b Ancoli-Israel (112) 4C-b Ancoli Israel (159) 5D-b Baker (191) 4C-b (151) Berger 4C-b (148) Berger 4C-b Table 5— Table AuthorCitation - / Level Evidence Alessi (162) 4C-b Alessi (163) 5D-b

SLEEP, Vol. 26, No.3, 2003 382 Actigraphy Review Paper—Ancoli-Israel et al Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Cordoba (154) Unblinded, nonrandomized, NS (40) / Grp1 51 (37-69), Questionnaire / Actillume / ACT: TIB, SE, num wakes, NS / Shift work, alcohol / Compared to normals: motor observational study, case-con- Grp2 50 (NS), Grp3 52±2, Wrist / 1 min sampled 20 sec WASO Low sample in the sub-group activity ↓ in pts, fragmented 4C-b trol study / Home / Entire 24- Grp4 52±2 / Cirrhosis / NS / Action 3 of ACT pts sleep ↑ in pts, rhythm damp- hs; ACT 5D, M-F (no week- ened in pts ends) Corkum (167) Unblinded, nonrandomized, 50 (50) / Grp1 9.1±1. 4 (7- Child Sleep Questionnaire- log; ACT: total sleep dur, NS / verbal IQ & perform- CSQP: ADHD Grp had ACT results were not com- observational study / Home / 11), Grp2 9.7±1.3 (7-11) / Parent version (CSQ-P); log SOn, num Nt awakenings, ance IQ<80, brain injury, per- longer sleep durs, more diffi- pared to CSQ-P or log find- 5D-b Entire 24-hs; 7D ACT Grp 1 ADHD, Grp2 Normal completed by parent / MML / restlessness; CSQ-P: SOn dif- vasive developmental disor- culty w/ SOn; more ADHD ings controls ndom wrist / NS / >.01g / ficulties, Nt waking, difficul- der, autism, psychosis, PTSD, subjects were “restless sleep- ActionW2 ties arising, restlessness, BT primary disorder of anxiety or ers” had more BT resistance; resistance affect / NS ACT: no statistically significant differences between the two Grps; log: ADHD Grp had longer sleep dur & more BT resistance

Cruise (165) Observational, prospective 276 (225) / 84.9±9.8 / NS / NS / wrist / 2 mins / NS ACT: TST, % sleep, peak >65 yrs, incontinence, no in 42% of waking episodes were study / NS / Nocturnal only Incontinence / NS sleep, mean dur; body move- dwelling catheter / NS / None associated w/ noise, light or 4C-b for 10 hs; 2Nt (non-consecu- ment incontinence care tive)

Dagan (192) Unblended, nonrandomized, 24 (24) / Grp 1 M, 12; Grp 2 Questionnaire (parent) / ACT; SE, Activity level, Inclusion: Grp 1 =ADHD. ADHS children had lower Subjective parental reports of observational study /home / M, 12 / Grp1, 1 9.6 Actigraph – AMI / hand / NS ‘quiet’ sleep percentage, Grp 2, healthy children / SC, higher activity level, and sleep not compared to ACT 4C-b nocturnal only for NS h. 3 Nt (6-12, SD = 1.6) Grp 2, 7.9 / NS / automatic scoring SON, Sdur; Others: S/W small sample size. Controls lower ‘quiet’ sleep percentage ACT (6-9.6, SD = 1.2) / Grp 1 = analysis program (CAMI) questionnaire variables NS significantly younger than than controls. No differences ADHD cases between groups on subjective parental reports of sleep.

383 Daurat (97) Controlled clinical trial; DB / 36 (24) / 51.2±2.2 / Normal Temp (n=19) / Actiwatch / Temp; ACT: sleep dur, activi- Healthy, no meds x3 mos, no Zopiclone improves ACT ACT detected both circadian Hotel rooms set up as lab / ndom wrist / 1 min / NS / ty index, ACT acrophase & time zone travel / Core temp identified sleep after trans- shift and sleep disturbance 4C-b Entire 24-hs; 6D ACT BL, 1D Actisom (P.Denise, France) amplitude; Mood: VAS not unmasked meridian travel F(1,22)=6.3, temp BL, flight, 6D ACT post –sleep visually scored, cosi- “mood”; Jet lag Sx VAS p<.05; The greater the dysyn- flight, 1+1D temp post flight, nor for phase “tired”, “ill-being”, “diges- chronization of CRs post- Zopiclone or Placebo before tive”, “energy” flight the shorter the sleep bed Nts 1-4 post flight dur. Zopiclone did not improve SBJ jet lag

Dinges (193) Controlled clinical trial, 20 (16) / 22.9 (NS) / Normal MSLT on D 2 of BL & D 5 of log; ACT; Mood: POMS; NS / PSG screening to ensure ACT data not specifically No specific ACT data report- unblinded, nonrandomized, sleep restriction; SSS; log w/ PVT, PRMT they had no sleep disorders, reported. Agreement between ed. 5D-b subjects served as their own VAS of SQ and several other MSLT to ensure their daytime ACT & other sleep measures controls / Laboratory (in the measures; Observation by sleep propensities were was NS. Univ. Gen. Clin. Res. Ctr) / others (none were compared between 8 & 20 mins / low Entire 24-hs; 10-12D ACT to ACT) / NS / wrist / NS / sample size

Actigraphy Review Paper—Ancoli-Israel etal consecutive (2D BL, 7-8D NS / NS sleep restriction, 1-2D recovery sleep)

Duka (139) Controlled clinical trial, DB, 8 (8) / 27.1±0.5 (21-34) / PSG (Nihon-Kohden 17- Endocrine measures-plasma NS / Extreme morning & B-carboline ZK 93426 (ben- ACT was not compared to the nonrandomized, cross-over Healthy channel, EEG, EOG, EMG); prolactin, GH cortisol, ACT: evening types, no acute or zodiazepine antagonist w/ other sleep measures 5D-a design, placebo-controlled / Temp (rectal); VAS / frequency of movement, chronic illnesses, no meds for weak inverse agonist activity) Laboratory / Nocturnal only Actometer ZAK-GMBH / intensity of movement; VAS 4wks, no h/o sleep distur- induces activation measured for NS h; 2Nt ACT (1Nt ACT dom wrist / 2 mins / >.1G bances, no benzodiazepines / by ACT & disturbs sleep placebo, 14D washout, 1Nt Low sample size measured by PSG ACT Tx or 1Nt ACT Tx, 14D washout, 1Nt ACT placebo) Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Dursun (152) Unblinded, nonrandomized, 24 (24) / Grp1 36.1±9.2, Grp2 VAS /Actigraph / Rt wrist / ACT: movement index (MI)= NS / Rxs that affect 5-HT Nighttime MI in Grp2 > Grp1 VAS results not directly com- observational study / Home / 35.1±11.5, Grp3 33.8±11.6 / NS / NS / Actplan and Actstat total num of points w/ move- concentrations; chronic alco- but no significant differences pared to ACT 5D-a Nocturnal only for NS h; 5Nt Grp1: schizophrenia on software (Switzerland) ment ÷ total num of points holism; CNS infections; can- between Grps1 & 2 in VAS ACT risperidone; Grp2: schizo- recorded; VAS rating of SQ cer; medical disorder that scores phrenia on a “typical” and morning sleepiness causes sleep abnormality / antipsychotic (not risperi- low sample size; not random- done); Grp3: normal control; ized Grps1&2: DSM IV Dx of schizophrenia for at least 2 yrs.

French (140) Unblinded, nonrandomized, NS (32) / NS / Normal None / NS / wrist / NS / NS / Melatonin: Salivary; Operationally qualified crew Crews had less sleep during observational study / High Walter Reed Algorithm Endocrine: Salivary cortisol; members participating in sim- mission 1, compared to mis- 4C-b fidelity weapons systems (General Activity Program) Mood: POMS; body pain sur- ulated, long dur bomber missions 2 & 3. Recovered nor- trainers (flight simulators) / vey; cognitive performance sions / NS / NS mal sleep dur w/in 48 hs after Entire 24 hs; 12D ACT test battery, activity log, 2- mission 3. Least restful sleep mins electrophysiologic after mission 1. measure, 30-S voice record, oral temp, fatigue score Friedman (136) RCT, SB, parallel design, 39 (35) / Grp1 61.9±7.1, Grp2 PSG; MSLT; SSS; log; urine Log: TST, TIB, SE, SL, NS / Acute unstable medical Few between-Grp differences ACT correlated more highly ACT compared w/ other tech- 65.1±8.6, Grp3 65.1±5.8 / samples to ensure no sleep WASO; PSG: TIB, TST, SE, or psychiatric illness, chronic in Tx efficacy. than log w/ PSG 2B niques in a subsample / NS / Insomnia, Age ≥55 med use / (AMI) / ndom wrist SL, WASO; ACT: TST, SE, illness associated w/ insom- Entire 24-hs; 4D ACT BL, 4D / 30 sec / NS / ACTION 1.32 SL, WASO nia, specific sleep disorder ACT Tx, 4D ACT f/u such as sleep apnea, sedating or stimulating meds, no sleep meds for ≥3 wks / NS

Friedman (126) Unblinded, nonrandomized, 101 (48) / 70±6.9 (56-88) / None / Actigraph, Ambulatory ACT: TST, SE, SOn, WASO, Participants in longitudinal Significant correlations Measures of poor sleep by 384 observational study, cohort Alzheimer’s disease (AD) by Monitoring Inc / wrist / 30 circadian activity, (amplitude, study of AD / NS / Non- between higher Time Based ACT correlated w/ more over- 4C-b study / Home / Entire 24-hs; NINCDS-ARDRDA criteria sec / NS /ACTION 1.3 acrophase, mesor); response bias - ACT data in Behavioral Disturbance all disruptive behavior in AD; 6D ACT Behavioral: MMSE, Time only 48 “records” from sam- Questionnaire overall score, but not w/ nocturnal disrup- Based Behavioral Disturbance ple of 101 subjects SE & WASO. No significant tive behavior Questionnaire, Alzheimer’s correlation between circadian Disease Assessment Scale activity measures and Time Based Behavioral Disturbance Questionnaire

Garfinkel (135) RCT, DB, cross-over design, NS (12) / 76±8 (68-93) / Mel / Somnitor (Neurim ACT: SL, SE, TST, WASO NS / NS / Low sample size SE was significantly greater Urine 6- sulphatoxy-mel placebo-controlled / Home / Insomnia; Independently liv- Pharmaceuticals, Tel Aviv, after w/ mel than placebo and measured at BL only 4C-b Nocturnal only for NS h; 3Nt ing older people w/ medical Israel) / wrist / NS / NS / WASO was shorter w/ mel ACT BL, 3Nt ACT Tx, 1 wk conditions and w/ other meds. “automatic scoring algorithm” than placebo washout, 3Nt ACT Tx by Cole, Kripke et al

Actigraphy Review Paper—Ancoli-Israel etal Gertner (171) Randomized, prospective / 36 (34) / M, 20, F, 14, ( new- None / AMI minature acti- ACT; % quiet sleep, activity NS/NS Early biological maturity Used in preterm newborns out of lab, hospital bed / 72 h, borns) / normal & premature graphs / ankle / NS /NS / level, TST related to child’s developmen- 4C-b 2 successive sessions infants Sadeh tal status. Decreased TST, increased activity at night all predictive of higher developmental scores

Glod (168) Unblinded, nonrandomized, NS (44) / Grp1 8.3±1.9, Grp2 None / Motion logger (AMI) / ACT: TST, SE, SOn, TIB, NS / Grp1 no significant psy- Controls vs. depressed chil- Waist placement of the acti- observational study, case-con- 9.4±2.3, Grp3 10.0±1.6 / waist w/ wrist in a subsample WASO, num nighttime awak- chopathology / NS dren had no significant differ- graphs 4C-b trol study / Home / Entire 24- Grp1 normal, Grp2 physical / 5 mins / Accelerations >.01g enings; Behavioral: Child ences in sleep parameters by hs; 3D ACT & or sexual abuse, Grp3 / Mactivity, Activity Analyze behavior checklist; Mood: K- ACT. Physical abuse major depression or dys- SADS-E (Schedule for appeared to be the factor thymia Affective Disorders and associated w/ sleep distur- Schizophrenia for School-age bance rather than Post Children-Epidemiologic ver- Traumatic Stress Disorder. sion) Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Glod (194) Unblended, randomized, non- NS / (22) Grp 1 M – 13, F – None / NS / belt-worn / NS / ACT: diurnal activity, diurnal Inclusion: medication-free Abused children were 10% randomized, observational 6; Grp 2, M – 9, F-6 / grp 1, NS / NS skew, % low-level activity, and medically healthy (two in more active than normals 4C-b study, case-control study / 9.4 (SD = 2.3); Grp 2 8.3 mesor, amplitude, % relative Grp 1 had illness) / NS / (p<.05), and had few periods home or hospital inpatients / (SD = 1.9) /none, Grp 1 = amplitude, acrophase; Mood, Patient selection (Grp 1 most- of low-level daytime activity entire 24 h, 3D ACT physically and or sexually schedule for Affective ly in patients, Grp 2 outpa- (p<.01). Later age of abuse (weekdays) abused; Grp 2, healthy con- Disorders and Schizophrenia tients) was associated with circadian trols for School-Age Children- dysregulation. Epidemiologic Version; Child Behavior checklist Haimov (195) Controlled clinical trial, DB, NS (51) / Grp 1; M, 4, F, 4; Questionnaire, sleep log, Sleep diary, log , melatonin, Inclusion: Participants in Improved SE and activity randomized, cross-over Grp 2, M, 6, F, 12; Grp 3, M, diary, melatonin / NS / wrist / ACT; TST, SE, SL, mean prior study of mel and sleep, level during sleep with a 2 mg 4C-b design / home (33 subjects), 19, F, 6 / Grp 1; 73.1 NS /NS /Sadeh, et al. activity level during sleep Grp 1 and Grp 2 had lower sustained release mel, and NH (18 subjects) / entire 24 (SD=3.9); Grp 2; 81.1 (SD = peak mel secretion / NS / NS improved SL with 2 mg fast- h, 7 D ACT; 7 D ACT Tx 1, 8.9) Grp 3; 71.4 (SD = 5.2) / release mel. Further improve- Tx 2, and Tx 3: 7 D ACT dur- other: Grp 1 independently ment after 2 mo 1 mg sus- ing last week of Tx 4 (2 mo); living insomniacs; Grp 2, tained release mel. 7 D ACT f/u (Grp 2 only) Tx institutionalized insomniacs; = melatonin (2 mg fast or Grp 3, independently living sustained release or 1 mg sus- elderly w/o sleep disorders, tained release) or placebo/ good clinical condition, not depressed, not demented

Hatonen (117) SB, nonrandomized, compar- NS (5) / 14.6 (12-19) / log / (AMI) / wrist / 1 min / log; ACT: normal vs. frag- All had NCL: / NS / low sam- 2 subjects had abnormally Very little actual data report- ative Tx, placebo-controlled / Insomnia, neuronal ceroid NS / Action 3.15 mented motor activity rhythm ple size fragmented activity rhythms ed. 5D-b Home / Entire 24-hs; 7D ACT lipofuscinosis (NCL) (period analysis by max during BL & placebo or mel BL, 7D ACT Tx w/ placebo, entropy spectral, auto-correla- didn’t affect rest/activity 7D ACT Tx w/ 2.5mg mel, tion and harmonic analyses) rhythms. In 3 subjects fami-

385 7D ACT Tx w/ 5mg mel lies reported slightly improved SQ w/ mel.

Hindmarch (141) RCT, unblinded, cross-over NS (30) / 27.3±0.7 (19-36) / LSEQ / AMI AMA-32 ACT: TST; Critical Flicker Healthy nonsmokers / NS / Caffeine had dose dependent design, each subject acted as Healthy habitual tea & coffee Motionloggers (AMI) / ndom Fusion Test; Choice Reaction Subjects were not blinded to negative effect on getting to 4C-b their own control / Laboratory drinkers wrist / 30 sec / NS / Stanley time; Line Analogue Rating the drinks sleep and quality of sleep by - Medical Research Center / 1997 Scale LSEQ and dose-dependent Entire 24-hs; 5 separate test negative effect on TST by Ds of ACT w/ at least 6D ACT washout period between test Ds

Hindmarch (142) RCT, DB, crossover design / NS (24) / 32.6 (19-58) / None / Ambulatory ACT: % sleep, % W; Critical NS / No meds other than Promethazine (an antihista- Laboratory - Medical Healthy normal volunteers Monitoring, Inc / ndom wrist Flicker Fusion Threshold; OCP’s, no alcohol, nicotine or mine w/ sedating side effects) 4C-b Research Ctr. / Entire 24-hs; / NS / NS / ACTION 3 Choice Reaction Time; Line caffeine / NS increase % sleep during the 6D ACT Tx, Washout of 4D Analogue Rating Scales for daytime and across the study Actigraphy Review Paper—Ancoli-Israel etal or more between test Ds Sedation period compared w/ fexofenadine, loratoadine and placebo.

Humphreys (196) Observational study / 54 (50) NS (NS) / battered Sleep log-diary/mini-motion Sleep diary-log – behavioral; NS/NS/NS Battered women experience ACT sleep assumed to be true women’s shelter / interview women logger / wrist / 30 sec / NS / Pittsburgh Sleep Quality sleep disturbance and daytime sleep 5D-a and 2D ACT & log Action 3 Index fatigue, VAS; ACT fatigue

Ito (197) Controlled clinical trial, NS (30) M, 12, F, 16 / 78.3 None / AMI / NS / NS / NS / ACT: D and NT percent NS/NS/NS No change with bright light unblended, nonrandomized, (NS) / normal, Alzheimer’s NS/ sleep, D and Nt awakenings, alone. Bright light plus vita- 4C-b parallel design / NS / entire dementia Others, MMSE, Clinical min B12 decreased daytime 24 h, 7 D ACT BL, 7 D ACT Dementia Rating percent sleep and num of Tx 1 (bright light) 7 D ACT naps. No Nt differences Tx 2 (bright light or bright light and vitamin B12 Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Jean-Louis (143) Unblinded, nonrandomized, NS (273) / M 51±7, F 52±7 None / Actillume (AMI) / ACT: TIB, TST, SOL, SE, NS / NS / Comorbidity NS Significant gender differ- Authors refer to their unpub- observational study, cross- (All 40-64) / Community wrist / 1 min / NS / Action3 sleep (amplitude, mesor, and ences, men had shorter TST, lished work comparing wrist 4C-b sectional study / Home / dwelling residents of San (Cole et al 1992 w/ Webster’s phase), activity (amplitude, lower SE, shorter TIB, lower ACT to PSG in women aged Entire 24-hurs; 3D ACT Diego identified by random re-scoring rules) mesor, and phase), illumina- Sleep amplitude, higher illu- 51-77 w/ 89% min-by-min telephone survey. tion (amplitude, mesor, and mination amplitude and high- agreement, r =.90. phase); SBJ mood er illumination mesor; significant white vs. minority differences, white sample had longer TST, longer SE, shorter SOL, greater sleep amplitude, and higher sleep mesor. Total sample average TST was 6.22 hs. Jean-Louis (122) Unblinded, nonrandomized, NS (32) / 44.76±20.64 / SSS, log / Gaehwiler ACT: TST, SE index, WASO, NS / NS / Small sample size Women had a better sleep Relatively small sample observational study / Home / Healthy normal volunteers Electronics (Hombrechtikon, SOL, frequency of transitions profile than men by ACT 4C-b Entire 24-hs; 5D ACT Switzerland) / dom wrist / 60 between sleep and wakeful- sec / ≥.1g / ADAS ness, daytime activity level, auto- r (amplitude of activity) Jean-Louis (46) Unblinded, nonrandomized, 273 (273) / (40-64) / Log / Actillume (AMI) / 1 ACT: TST, SOL, SE index, Community dwelling resi- CR of illumination was sig- Higher amplitude of activity observational study, cross- min / NS / NS / Automatic mesor, amplitude (of the dents of San Diego identified nificantly associated w/ activ- (by ACT) was associated w/ 4C-a sectional study / Home / scoring rhythm; ACTION3 cosine) and phase (timing of by random telephone survey / ity and sleep rhythm meas- less reported daytime nap- Entire 24-hs; 3D ACT the peak of the fitted cosine) NS / NS ures. Higher amplitude of log ping. (r=-.16, p<.05). Level of illumination illumination correlated with sleep phase (r = .16), lower SE index (r=-.15), and less reported daytime napping (r=- .16). Higher amplitude of 386 activity correlated with sleep amplitude (r=.30), Sdur (r=.21) and less daytime napping (r = -.16). Jean-Louis (198) Unblinded, nonrandomized, NS (24) M 17, F, 7 / 45 (SD PSG / Gaehwiler / writ / 60 PSG: SOL, WASO, SE, TST; Inclusion; recruited from ACT scoring criteria based on ½ of sample used to calibrate observational study / labora- = 9) / 18 with insomnia, 6 sec / >0.1g / ADAS ACT: TST, SE, Mood: Beck ongoing PSG studies / NS healthy young adults did not new ACT scoring algorithm 2B tory / nocturnal only for NS with hypersomnia, current Depression Inventory, HAMD /NS perform as well in this and other ½ of sample used to h, 2 Nt PSG, 1 Nt ACT (on major depressive episode depressed sample as criteria validate the new algorithm second NT of PSG) optimized for this sample. Jean-Louis (199) DB, randomized, cross-over NS (10) / M, 4, F, 6 / 68.8 SSS, sleep log, diary, VAS / Sleep diary, log, ACT; TST, Inclusion: Aging and Mel enhanced rest-activity design / home / entire 24 h, (SD = 15.8) / self-reported NS / wrist / NS / NS / ADAS SE index, WASO, TWT, SWT Alzheimer’s Registry, self- rhythm, reduced SOL, 4C-b 5D ACT Tx 1 (6 mg mel or s/w disturbance, Alzheimer’s and rest-activity amplitude; reported s / w disturbance,/ reduced SWT, and improved placebo); 5D ACT Tx 2 disease Others: Alzheimer’s Disease Exclusion, sleep apnea, PLM, recall of previously learned

Actigraphy Review Paper—Ancoli-Israel etal (placebo or 6 mg mel) Assessment Scale, Digit span, medical, neurological or psy- materials Digit symbol substitution, fin- chiatric illness other than ger tapping, MMSE Alzheimer’s disease / low sample size

Kario (200) Unblended, nonrandomized, NS / (231) M, 126, F, 105 / None / AMI / waist while ACT: physical activity during Inclusion: enrolled in larger Depression was associated Nt physical activity (by ACT) observational / NS / 1 D 46 (30-66, SD= 8.9) / normal awake, wrist during sleep / S and W, Mood; Brief work site blood pressure with disrupted diurnal BP was used to estimate SQ, but 5D-b simultaneous ACT and ambu- NS / NS / NS Symptom Inventory ) depres- study/ Exclusion no hyperten- variation independent of no ACT sleep findings report- latory BP monitoring sion and anxiety) Symptom sion, depression or anxiety physical activity by ACT in ed rating checklist-90-Revised; meds; not shift worker / NS men, but not women Others: 24 h ambulatory BP monitoring

Lee (201) Cross-sectional study / Home 104 (100) / 38.3±7.8 / HIV Questionnaire / MML / wrist / Log; ACT; Mood NS / Dementia neuropathy Pts averaged 6.5 h/Nt; 45% / Entire 24 h; 2 D ACT NS / NS / Action3 use of illicit substances, hos- napped. CD4 cell counts 4C-b pitalization / NS unrelated to sleep, high fatigue grp had more disturbed sleep and more depression symptoms. Rhythm strength inversely related to daytime dysfunction, directly related to napping, TST. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Lemke (153) Unblinded, nonrandomized, NS (52) / 43.0±12.5 / Questionnaire, log PSQI / Log: TST, SWT, SQ VAS; NS / Psychotic features and PSQI “poor” sleepers had Manuscript Table 2 incorrect- observational study / Entire Psychiatric unit in-pts w/ Aktometer (ZAK, Germany); ACT: mean activity; Mood: or neuroleptics, meds changed higher mean nighttime motor ly labels poor & good sleep- 4C-b 24-hs; 3D ACT DSM IV criteria for major ndom hand; 2 mins; HAMD (German versions) during study period, motor activity than “good” sleepers; ers (correct in the text); ACT depressive disorder Sensitivity: >.1g; Scoring: system disorders (e.g. Depressed pts w/ less depres- data only analyzed as mean ZAK software Parkinson’s disease) / Low sive symptoms had lower activity levels, not S/W sample size of good sleepers mean nighttime motor activity for comparison to poor sleep- & higher log-reported SQ ers by PSQI, differential reporting bias (more severely depressed pts may exaggerate sleep problems on log)

Maus (137) Controlled clinical trial, NS (NS) / Grp1 31 (22-45), None /Actigraph (AMI) / NS Endocrine: urine norepi and NS / h/o eye surgery, chronic Sleep apnea Grp had higher unblinded, nonrandomized / Grp2 56 (38-74) / Normal in / 1 min / NS / Sadeh, epi; ACT: Activity index, SEI eye disease, asymmetric nighttime activity index by 4C-b Laboratory / 22 hs; Grp1 Grp1 (healthy controls-stu- Technician Sleep Laboratory, Nighttime pulse oximetry- intraocular pressures, meds ACT and lower SEI compared (normals) 2D ACT (1D ACT dents & health care workers), Israel desats known to affect aqueous to controls allowed to sleep normal control, 1D ACT no severe OSAS in Grp2 humor flow / Patient selec- sleep allowed), Grp2 (Sleep tion: Grp2 older than Grp1 Apnea) 1D ACT allowed to sleep but w/o respiratory assistance (nasal CPAP)

McArthur (169) RCT, DB, Tx order, cross- NS (9) / 10.1±1.5 / Rett log by caregivers, SOL, num log; ACT: TST, SE, num of NS / NS / Low sample size SOL significantly reduced by High variability in subject over design, comparative Tx, Syndrome (stage III or IV) of awakenings & final morn- awakenings, SOL mel during the 1st 3 wks of responsiveness to mel 4C-b placebo-controlled / Home / ing awakening / (Ambulatory trial Entire 24-hs; 10 wks of ACT Monitoring, Inc) / NS / NS / (7D ACT BL, 28D ACT Tx, NS / Action3 7D ACT washout, 28D ACT

387 Tx)

McCarten (202) Controlled clinical trial, NS ( 7) / M 7, 73 ( 62-81) / observation by nurses / AMI / ACT: SO, # awakenings, TST Alcohol or drug abuse, seri- Nurses observations agreed unblinded, nonrandomized, disrupted sleep, Alzheimers dom wrist / NS / NS / NS during day , mean activity ous medical illness, severe with actigraphy. Triazolam 4C-b cross-over design / lab / 10 D Disease depression / low sample size had no significant effect on & N ACT TST at night SO, # arousals, or TST during day. No sig drug effects on memory

Mendlowicz (144) Unblinded, nonrandomized, NS (32) / 50.00±23.97 (18- SSS, log / Gaehwiler ACT: daytime activity level, No psychiatric or medical Age was associated w/ TIB, observational study / Home / 79) / Normal community Electronics / wrist / NS / NS / TST, SOL, transition from problems / NS / small sample TST & WASO; significant 5D-a Entire 24-hs; 5D ACT dwelling volunteers ADAS sleep to wakefulness, WASO, predictors of depressed mood TIB; Mood: depressed mood were daytime activity level, from Alzheimer’s Disease SOL, WASO, TST, & TIB Assessment Scale; Alzheimer’s Disease

Actigraphy Review Paper—Ancoli-Israel etal Assessment Scale, SSS, MMSE, VAS Mennella (170) RCT, SB, cross-over design / 15 mother/infant pairs (13 None / AMA-32 (AMI) / ACT: sleep %, total mins Nonsmoking, lactating Short-term exposure to small Authors refer to another study Private carpeted room w/crib mother/infant pairs) / Grp1 infant’s Lt leg / 1 min / NS / quiet sleep, total mins active women and their infants / NS amounts of alcohol in breast where adding vanilla flavor 4C-b / 3.5 hs; 2 test Ds in each 27.4yrs±1.1 (22-34), Grp2 Sadeh et al sleep, longest SP, SL, num of / NS milk alters sleep in infants did not alter the infant’s sleep infant, separated by 1 wk; 2.7mos±0.3 (1.5-5.6mos) / sleeping bouts, mean activity infants bottle-fed 100 ml of NS count during wakefulness breast milk w/ 40ml (32mg) ethanol

Miaskowski (149) Unblinded, nonrandomized, NS (22) / 56.6±13.0 / Cancer Log; Likert scales of SQ and ACT: TST, SE & num of Receiving radiation therapy No significant correlations No evidence of first night observational study / Home / pts w/ painful bony metas- feeling rested / MML / wrist / awakenings; Mood: CES-D; for painful bony metastases, between the OBJ measures of effect b/t 2 consecutive nights 4C-b Entire 24-hs; 2D ACT tases NS / NS / Action3 Others: Lee Fatigue Scale, age ≥ 18, spoke English, sleep and self-ratings of feel- of wrist ACT pain numeric rating scale Karnofsky score ≥ 50, pain ing rested quality of sleep. intensity ≥ 2.5 on a 0-10 Time spent napping was posi- scale, x-ray bone mets & on tively related to increased opioids / NS / NS amounts of sleep the night before. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Miller (203) Unblended, nonrandomized, NS (15) Grp 1 None / NS /NS /NS /NS/ NS ACT: motor activity “ while NS/NS/low sample size Wake activity, sleep activity Minimal specific information case series / home / entire 24 methylphenidate, M-6; F-2; awake” and “while asleep”, and hyperactivity decreased on the study’s methodology 5D-a h, ACT BL, ACT placebo, Grp 2 Pemoline, M-5, F-3 / Others: Connors parents’ rat- with methylphenidate. Awake and the lack of statistical test- ACT Tx (with Grp 1; 7.5 (SD = 2.1) Grp 2 ing scale (hyperactivity); activity increased and asleep ing reported makes the find- methylphenidate or pemo- 9.1 (SD = 2.3)/ normal, Global rating scale (of activity and hyperactivity ings difficult to interpret. line), num days ACT NS ADHD, Grp 1 received improvement) index decreased with pemo- methylphenidate, Grp 2 line (statistical testing NS) received pemoline

Mishima (99) RCT, unblinded, crossover 22 (NS) / Grp1 81, Grp2 78 / None / (AMI) / ndom wrist / ACT: Total, daytime & night- DSM IV Dx of vascular Vascular Dementia: daily Shows ACT detected Tx design / NH / Entire 24-hs; Grp1 Vascular dementia, 1 min / NS / activity counts time activity, % night/total dementia or Alzheimer’s / bright light reduces nighttime effect in well-designed, con- 4C-b continuous ACT for 1 wk pre- Grp2 Alzheimer’s dementia used, sleep not scored activity mixed dementia excluded / activity; Alzheimer’s trolled trial. Suggest ACT Tx, 2 wks Tx, 1 wk post-Tx, Unable to blind Tx, small N Dementia: bright light does useful to detect outcome in ≥ 4 wks washout cross-over; not affect activity rhythm CR Tx study Tx=5-8000 lux 9-11am; Control=300 lux 9-11am

Moorcroft (145) Unblinded, nonrandomized, NS (15) / (19-62) / People log to get intended W time for log; ACT: Sleep & W periods, Able to self-awaken / NS / Data suggested that these observational study / Home / able to self-awaken at a self- the next morning /Actigraph time of final awakening Low sample size people could awaken at a self- 5D-b Nocturnal only from 8:00pm predetermined time w/o exter- (AMI) / NS / NS / NS / NS determined time. until 10 mins after awaken- nal means ing; 3Nt ACT

388 bed & out of bed, mean activ- ed less fatigue. Rest/activity ity; Mood: HADS (scale of rhythm remained a significant anxiety and depression); predictor of survival in multi- QLQ-C30 (QOL) variate analysis. Ouslander (166) Case series, descriptive / NH / 73 (73) / 86.9±7.2 / NS / NS / NS / 2 mins / NS / ACT: Sleep intervals NS / NS / No information on Incontinence was not related Built own actigraph Nocturnal only for 9 hs; ACT Incontinence NS how (sleep) ACT recorded to sleep disruption 5D-b 1 Nt BL, 2 mos later, ACT 1 Nt 2nd BL

Pankhurst (146) Unblinded, nonrandomized, Study1 96 (92) / (23-67) / Log / Gaehwiler Electronics log; ACT:”actiblip” inci- NS / Hypnotics, pain that Men had more movements. ACT data not scored as S/W observational study, cohort couples slept together, Hombrechtikon, Switzerland / dences seriously disrupted sleep / For sleep partners, 1/3 of but as “actiblips” (movement) 4C-b study, cross-sectional study / Study2 (95) / NS / subjects NS / 30 sec / >.1g / Comorbidity was not movements were common to Home / Nocturnal for NS hs; who slept alone. All subjects ACCORD (author’s own pro- addressed both partners; subjects sleep- 8Nt ACT were community dwelling gram) ing w/ partner had more people living near 4 UK air- movements; movements

Actigraphy Review Paper—Ancoli-Israel etal ports decreased when partner absent

Pat-Horenczyk (160) Unblinded, nonrandomized, 77 (67) / Grp1 85, Grp2 87, None / Actillume / NS / NS / ACT: mins asleep each hr, NS / None (most pts had mul- SDG slept more at each hr, observational study / NH / All 85±7.3 (60-100) / NH res- ≥.003g / NS fully asleep each hr, fully tiple medical diagnoses and and more than during meal 4C-b Entire 24-hs; 1D ACT idents w/ dementia; SDG- awake each hr were taking multiple meds) / times. Peak of sleep occurred MMSE<20, MMDG-MMSE NS 2 hs earlier in SDG Grp, ≥20 MMDG had more wakefulness at night and SDG had more sleepiness during D

Paul (204) Unblended, nonrandomized, NS (NS) NS (NS) / normal None / Precision control ACT: num and dur of sleep Inclusion: Canadian Air Amount of daily sleep observational study /home / Design / wrist / 1 minute / NS episodes; Others, psychomo- Command pilots and co-pilots decreased during the 3 D 4C-b entire 24 h, ACT 5 D prior to / Win Act 1.2 tor tasks; serial choice reac- / NS/ NS prior to mission. During mis- re-supply mission and until tion time, logical reasoning sion, worst sleep on night 2. mission completed task serial subtraction task During mission, decreased and multitask self-rated alertness and increased fatigue Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Pollak (138) Case-control study / Home / 15 (15) / Grp1 75.9, Grp2 None / Vitalog PMS-8 / wrist log; ACT: activity NS / NS / Small sample size Rx uses (i.e. those taking Entire 24-hs; 12 Nt&D ACT 72.0 / NS / 1 min / NS / NS meds at BT) become active 4C-b 1.5 hs earlier in the morning

Pollak (158) Case series / Home / Entire 88 (44 pairs) / Grp1 77.9 (64- NS / MML / Ndom wrist / log; ACT: activity level 65+; have care giver; disrup- Behavior of elders & care- Presents a partial mathemati- 24-hs; 9 D&Nt ACT 92) Grp2 61.9 (29-86) / NS 0.5 min / NS / NS tive nocturnal behavior givers were less similar in cal model of spontaneous 4C-b (DNB) / NS / NS daytime than at Nt; at Nt it motor behavior. Used video was elders that initiated inter- at Nt to watch interactions action, thus disturbing sleep between elder & caregiver of caregiver

Redeker (156) Observational study; longitu- 22 (13) / Grp1 63.7±9.9 (43- NS / MML / ndom wrist / 1 ACT: TST, num wakes, mean NS / NS / Low sample in fol- Significant daytime sleep & Nice longitudinal follow-up dinal study / Home; out-of- 83), Grp2 62±10.76 (43-78) / min / NS / ACTION W time, mean sleep interval, low-up fragmented nighttime sleep (6 mos) 4C-b lab hospital bed / Entire 24- Coronary artery bypass graft % sleep; Sickness Impact during hospitalization; Over 6 hs; 7D postoperative in hospi- surgery Profile mos period post-surgery, tal, 7D Early recovery at sleep consolidated & daytime home, 7D resume activities, sleep lower; Perceived sleep 7D complete resumption of improvements consistent w/ activities ACT measures

Redeker (205) Case series, time series / Out- 25 (17) / Grp1 63.67±9.86 None / MML / ndom / 1-min ACT: Activity counts; NS / NS / Small sample Relationship between rhyth- Looked at rhythms as an out- of-lab hospital bed / Entire (42.83) / Coronary bypass / NS / Action Sickness Impact Profile mic & linear patterns of activ- come 4C-b 24-hs; Continuous for entire surgery ity w/ recovery; i.e. Positive hospital D relationship between length of stay & dysfunction w/ shorter activity periods associated w/ less dysfunction. Positive circadian activity periods were related to better functioning 389 & shorter stay

Redeker (157) Observational study, descrip- 40 (33) / M 56.1±11.9, F None / MML / ndom / 1-min / ACT: Night interval, TST, Exclusion: Known sleep dis- pre-hospital endogenous vari- tive, correlational / Out-of-lab 59.5±10.9 / Cardiac disease: NS / Action 3 SE, num wakes, dur of sleep orders, psychotropic meds, ables, age, gender, & New 4C-b hospital bed / Entire 24-hs; acute myocardial infarction, interval, dur of wakes; Veran blindness, deafness / NS York Heart Association func- Continuous for course of hos- unstable angina & Synder-Halpern Sleep tional classification and pre- pital stay (range = 1-10D) Scale-pts. Perception of their hosp sleep loss all related to sleep; Sleep loss subscale of SE & dur of wakes once hos- pt. Sleep screener pitalized. Pre-hosp severity of cardiac disease greatest correlate

Redeker (155) Case series / Out-of-lab hos- 22 (NS) / Grp1 57.12±6.6 None / MML / ndom / 1 min ACT: activity counts, cosinor Exclusion: H/o of psychiatric Activity levels & strength of Examined the effect of age pital bed / Entire 24-hs; (47-65), Grp2 72.36±4.1 (66- / NS / ACTION 3 parameters; Sickness Impact disorder, neuromuscular dis- CRs increased over Ds 2-5 4C-b Continuous for hospital stay 77) / Coronary artery bypass Profile order, chronic renal failure, post-operative. In older adults

Actigraphy Review Paper—Ancoli-Israel etal blindness, deafness / Small it took longer (at a slower Sample rate) Reyner (147) Unblinded, nonrandomized, NS (400) / Grp1 (20-34), log / “Swiss-type” Gaehwiler log; ACT: time of SOn, SQ door-to-door survey / sleep- Time of SOn was effected by observational study / Home / Grp2 (35-49), Grp3 (50-70) electronics / wrist / 30 sec / promoting meds or excessive gender & age; and differed 4C-b Nocturnal only for NS hs; / Sleep normal Community NS / Horne et al, “SOn algo- alcohol / NS from log and ACT; Average 15Nt ACT dwelling people living near 4 rithm” movement declined w/ age, UK airports and men had more movement than women

Sadeh (172) Observational., longtitudinal / 135/135 / M, 69, F, 66, NS None / Mini-motionlogger ACT; SO, morning up time, NS / NS Kids fragmented sleep had home / nocturnal, 5N (7.2-12.7) / normal, NS (AMI) / non-dom wrist / 1- total sleep period, % sleep, # lower performance and more 4C-b min / NS / Sadeh actigraphic awakes, longest sleep period, behavior problems scoring algorithm quiet sleep %, neuropsych tests Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Sadeh (206) Case-control study / Home / NS (NS) / Grp1 12.1 (8.2- None / AMA-32 (AMI) / ACT: SOL, TST, s1eep per- NS / NS / NS Asthma kids were more Was first home-based evalua- Entire 24 hs 3-4D 15.4), Grp2 12.4 (8.9-14.7) / ndom / 1 min / NS / ASA pro- cent, longest SP, % quiet active during sleep & had less tion of asthma & sleep in 4C-b asthma gram sleep; activity level; % quiet sleep; boys w/ asth- children Pulmonary function, asthma ma fell asleep faster than severity, sleep Q by child & girls; control girls fell asleep parent faster than boys; %sleep & mean activity correlated w/ evening & morning Pulmonary peak expiratory flow measures & poor sleep associated w/ more severe asthma

Sadeh (207) Observational study / out of 39 (NS) / Grp1 9.51±1.9 (7- None / AMA-32 / non-dom / ACT: SO, TSP, % sleep, true NS / NS / NS Lower SQ during hospitaliza- lab hospital bed / Entire 24 14) / Severe psychiatric prob- 1 min / NS / ASA sleep time, longest sleep per- tion strongly associated w/ hrs; 3D lems cent, % quiet sleep; WISC - self report on depression, 4C-b R, Depression scale, psycho- hopelessness & low self logical tests, Abuse history esteem in children w/ severe behavior disorders

Sakakibara (207) Unblinded; Randomized; 10 (10) / 59.7 (50-69) / NS / MML / ndom / NS / NS log; PSG; ACT: mean activi- Exclusion: On meds, mentally 60 mins of bright light in the cross-over design / Home / Normal / Cole et al ty, movement index, daytime ill, physically ill / Small sam- morning reduced nocturnal 4C-b Entire 24-hs; 5Nt ACT, PSG activity level ple activity last 2Nt

Scherder (114) Randomized / NS / Entire 24- 16 (15) / Grp1 81.7 (70-91) / NS / No info ACT: is (interdaily stability); Exclusion: Alcoholism, Psych TENS increased IS hs; 4D ACT BL 1, 4D ACT Alzheimer’s Disease IV interdaily variability; RA illness, cerebral trauma, 4C-b after 6wk Tx; 4D ACT after relative amplitude Cerebrovascular dis, Epilepsy,

390 6wks no Tx etc. / Not enough info

Schlesinger (208) Unblended, nonrandomized, NS (34) Grp 1, M, 4, F, 14; None / AMI / wrist / NS / NS Sleep diary, log; ACT, SL, Exclusion Grp 2: no meds, no No significant differences in observational study /home / Grp 2, M, 5, F, 11 / Grp 1, / Actigraph Statistical analysis total cumulative S dur, SE sleep problems, no h/o ACT measured sleep between 4C-b nocturnal only for NS h / 3 Nt 36.5, (20-48, SD = 8.5) Grp and num arousals whiplash injury whiplash pts and controls. In After whiplash injury and 3 2, 33.2, (20-50, SD = 8.0) , sleep logs, whiplash pts had Nt f/u (3-5 months later normal, Grp 1; acute whiplash prolonged SL, and impaired injury in traffic accident, Grp SQ. In whiplash pts, greater 2, normal controls signs/symptoms of whiplash correlated with greater num arousals and lower SE.

Schnelle (209) Controlled clinical trial, 267 (184) / No info / NS NS / NS ACT: Peak activity; mean Noise & light reduced corre- Measured noise levels, used unblinded, nonrandomized, activity lated w/ change in % sleep ACT only at Nt, used obser- 5D-b cross-over design / NH / but no change in D sleep (by vation of sleep during D Nocturnal for 10 hs; 5Nt ACT observation) or actual sleep Actigraphy Review Paper—Ancoli-Israel etal variables

Schwietzer (210) RCT; DB; comparative Tx / NS (12) / 31.3±8.1 / Normal; MSLT / (AMI) / NS / 1 min / Log: VAS; MSLT, simulated No ACT results reported. Not given any info on acti- Laboratory / 3D ACT Hypersensitivity to environ- NS / NS assembly line task after 3 Ds amount of sleepi- graph results 5D-b mental allergen ness reported and on MSLT, same in all Grps

Shamir (211) RCT; DB; cross-over design; 27 (19) / 42±5 (24-67) / NS / Somnitor / NS / NS / NS ACT: SE, SOL, TST, WASO, Exclusion: Liver or renal dis- SE improved w/ mel com- Analyzed those w/ poor sleep comparative Tx / NS / 3Nt Insomnia; Schizophrenia / NS Fragmentation index, num ease or other psych disorders, pared to placebo in those who to start out w/ separately 4C-b ACT during each Tx phase wakes severe physical illness / Much started out w/ poor sleep 2mg mel vs. placebo information missing on ACT

Shilo (212) Observational study / Out-of- NS (14) / Grp1 61±11 (39-76) NS / Somnitor / NS / NS / NS Melatonin; ACT: TST Very little sleep in a 72-h lab hospital bed / Entire 24- Grp2 59±11 (49-74) / All on / NS period 6-SMT abnormal in all 4C-b hs; 72 hs ACT ICU w/ no nocturnal rise

Shilo (213) Comparative Tx (mel/place- NS (8) / 62±14 (30-72) / On NS / Somnitor / ndom wrist / ACT: TST, num awakenings NS / NS / Small sample but Melatonin improved sleep- bo) / Out-of-lab hospital bed / pulmonary ICU-COPD 1min / NS / NS understated population both duration & quality 4C-b Entire 24-hs; 3D ACT Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No. 3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Shinkoda (214) Observational study / Home / NS (4) / (26-31) / Normal; NS / AMA-32CL (AMI) / log / ACT: TST, SE, SOL, NS / NS / Low sample but After delivery, sleep became Entire 24-hs; ~4 mos ACT pregnant or post-delivery ndom wrist / 1 min / NS / NS Naps, WASO unusual population irregular & WASO went up 4C-b

Shirota (215) Observational study / Home / NS (28) / Grp1 74.1±4.9, NS / NS / ndom / NS / NS / log / ACT: mean activity, NS / NS / NS Hi volitional aged pts had Entire 24-hs; 10-14D ACT Grp2 73.0±5.0 / Normal NS TST, SOL, TST-nap well structured sleep and 4C-b most took 1400h nap; low volitional had poor structured sleep & tool time-dependent nap ~ 8h after awakening

Stein (216) Controlled clinical trial; DB; NS (25) / 8±1.8 (6-12) / NS / NS / NS / NS / NS / NS log; ACT: activity level, SOL, NS / NS / NS Slept less when Rx given TIB cross-over design / Home / ADHD TST, num awake, duration of than in placebo 5D-b Nocturnal only for 16 hs; 2- wakes 18h periods during each of 4 wks of protocol Thessing (104) Cross-sectional study / 30 (24) / 21 (18-29) / Normal NS /(AMI) / wrist / 1 min / log; ACT: TST, WASO NS / NS / NS No difference in sleep Tested shift workers during Laboratory / Diurnal only for NS / NS between the different light the D after a night w/ bright 5D-b 8 hs; 1D ACT (while asleep conditions light Tx in lab during D)

Usui (217) ACT compared with other 18 (18) M, 11, F, 7/ 30.1 (NS) Sleep log, diary Sleep diary, log, ACT: sleep NS/NS/ low sample size Sensitivity (sleep) between Used ACT as gold standard to techniques/home/entire 24 h, normal /Motionlogger-AMI/ non-dom minutes, wake minutes ACT and log was 86.7% and validate sleep log in healthy 4C-a 5-7 D ACT & Logs wrist / 1 min /0.01 g/rad/s / specificity (wake) was adults NS 97.04%

Usui (218) ACT compared with other 35 (35) Grp 1; 27.6 , (NS) Sleep log, diary Sleep diary, log , ACT: sleep NS/NS/NS Sleep log does not always Used actigraphy as gold stan- techniques/home/ 2-7D ACT Grp 2; 74.3, (NS) Grp 3, /Motionlogger – AMI / non- & wake minutes detect sleep state. Agreement dard to validate sleep logs in

391 4C-a & log 42.1; (NS) normal, narcolep- dom wrist / 1 min / dropped during sleep-wake healthy vs. elderly vs. sleep sy, other; sleep state miscon- 0.01g/rad/s / Cole transition periods. Sensitivity disordered patients ception, idiopathic hypersom- was lower in patients with nia, DSPS sleep disorders than normals.

Usui (219) Unblended, randomized, 4 (3) / NS (64-80) / normal, Sleep log, diary, VAS-sleepi- Sleep diary, log; ACT; day NS/NS/ low sample size, no Decreased D and Nt TST (by cross-over design, case series NS ness / Motionlogger, AMI / TST, night TST; Others: sleep placebo ACT) with bright light. 5D-a / home / entire 24 h 3D ACT non-dom wrist / NS / .01 g / log variables: Son time, SOff Increased NT TST (by ACT BL, 4 D ACT Tx 1, 3 D ACT Cole et al time, day TST in maps with triazolam. Findings less- washout, 4 D ACT Tx 2. Tx = ened over the 4 D for both 2 h bright light 2500 lux at Tx. BT or triazolam .125 mg

Van Londen (220) Observational study / NS / NS (78) / Grp1 45.2±14.4 NS / Gaehwiler / ndom / 30 log / ACT: mean activity dur- NS / NS / NS During W: pts < mean activi- Entire 24-hs; 5D ACT (22-77), Grp2 42.2±16.3 (19- sec / NS / NS ing W & sleep; immobility ty than controls; pts lower 4C-b 73) / Depression during sleep; fragmentation fragmentation & lower TST; Actigraphy Review Paper—Ancoli-Israel etal During sleep: pts had higher activity Van Someren (115) Observational study / Home 42 (34) / 72±1.2 / NS / NS / wrist / NS / NS / ACT: IS (interdaily stability), NS / NS / NS Rhythm disturbance in most 18=Nt; NH 16=Nt / Entire Alzheimer’s EEPROM IV (intradaily variability) Alzheimer’s Disease pts, no 4C-b 24-hs; 155 hs ACT sundowning, rhythms in NH less stable than home particularly in IS Vercoulen (221) RCT; DB; parallel design / NS (96) / Grp1 38.5±10.1, NS / Actometer / ankle / NS / log; ACT: mean activity; Exclusion: Other psych disor- No difference in any outcome NS / Entire 24-hs; 12D ACT Grp2 39.9±8.6, Grp3 NS / NS Mood: Beck, SIP (SIP-sick- ders measures between the 4 Grps 5D-b 37.8±11.9, Grp4 39.8±7.4 / ness impact profile) Chronic fatigue syndrome; depression Wallace-Guy (222) Unblended, nonrandomized, NS (154) F, 154 / 66.7 (51- None / Actillume (AMI) / ACT: lux, in-bed sleep, out of Inclusion: participants in Greater 24-h illumination cor- observational study / home / 81) / normal, postmenopausal wrist / 1 min /10 sec / NS / bed sleep, SE, SL, wake with- womens’ health initiative related with shorter SL, 4C-b entire 24 h 7D ACT women NS in sleep, TST, sleep timing, observational study / reduced wake within sleep, amount of daytime napping; Exclusion/ physically unable and greater depressed mood. Mood: depression brief to participate or not likely to Evening light exposure was screening questionnaire survive few months / NS not associated with these variables. Citation - Author / Study Criteria Design / Sample Size (Completed Comparison Measures / Outcome Measures Inclusion / Exclusion / Bias Study Conclusion from Comments from Reviewer Evidence Level Location / Protocol Study) / Mean Age (Range) / Actigraph apparatus / paper SLEEP, Vol. 26,No.3, 2003 Medical Conditions Placement / Epoch Length / Sensitivity / Scoring

Wolter (223) RCT; DB; parallel design / 71 (NS) / 47.5 (23-66) / NS / (AMI) / wrist / 1min / log; ACT: mean activity, SE, Exclusion:OSA, EDS, PLMS, 1. NS diff in SE between dif- Home & out-of-lab hospital Smoker NS / (AMI) inact index during D narcolepsy, parasomnia / NS ferent patch doses; 2. At all 4C-b bed / Entire 24-hs; 7D ACT in phases of Tx mean act less hospital, 7D ACT last wk than BL patch Tx, 7D ACT off patch, 7D ACT @ 6mos f/u

Yesavage (224) Unblended, non-randomized, NS (42) 61.9% M, 38.1% F / None / AMI / non-dom wrist / Sleep diary, log, ACT : SE, Probably Alzheimer’s disease Between subjects differences observational, longitudinal/ 70.8 (SD = 7.5) / normal 30 sec / NS / Action software amplitude of rest, activity CR; and MMSE = 15./ major med- explained over 55% of vari- 4C-b home / entire 24 h 5D ACT version 1.32 Others; MMSE ical illness / NS ance over time in SE and CR. every 6 mo for an average 2.2 MMSE state explained only 5 yrs % of variance over time. Zhdanova (225) Case-control study / Home / 13 (NS) / 6.5 (2-10) / NS / mini-logger 2000 / pock- log; Endocrine Measures; NS / NS / Not blinded, no Sleep improved w/ mel Entire 24-hs; 7D ACT prior to Insomnia; Angelman syn- et on back of vest / 1min / NS ACT: num body movements, placebo control 4C-b mel. Tx, 5D ACT on mel drome / MIT algor. TSP 0.3mg 392 Actigraphy Review Paper—Ancoli-Israel etal PRACTICE PARAMETER Practice Parameters for the Use of Actigraphy in the Assessment of Sleep and Sleep Disorders: An Update for 2007 Standards of Practice Committee, American Academy of Sleep Medicine 1Timothy Morgenthaler, MD, 2Cathy Alessi, MD, 3Leah Friedman, PhD, 4Judith Owens, MD, 5Vishesh Kapur, MD 6Brian Boehlecke, MD, 7Terry Brown, DO, 8Andrew Chesson, Jr., MD, 9Jack Coleman, MD, 10Teoﬁlo Lee-Chiong, MD, 11Jeffrey Pancer, DDS, 12Todd J. Swick, MD

1Mayo Clinic, Rochester, MN; 2VA Greater Los Angeles Healthcare System-Sepulveda and University of California, Los Angeles; 3Stanford University School of Medicine, Stanford, CA; 4 Rhode Island Hospital, Providence, RI; 5University of Washington, Seattle, WA; 6University of North Carolina, Chapel Hill, NC; 7St. Joseph Memorial Hospital, Murphysboro, IL; 8LSU Health Sciences Center in Shreveport, Shreveport, LA; 9 Murfreesboro Medical Center, Murfreesboro, TN; 10National Jewish Medical and Research Center, Denver, CO; 11Toronto, Canada; 12Houston Sleep Center, Houston, TX;

Background: Actigraphy is increasingly used in sleep research and the with insomnia and hypersomnia, there is evidence to support the use of clinical care of patients with sleep and circadian rhythm abnormalities. actigraphy in the characterization of circadian rhythms and sleep patterns/ The following practice parameters update the previous practice param- disturbances. In assessing response to therapy, actigraphy has proven eters published in 2003 for the use of actigraphy in the study of sleep and useful as an outcome measure in patients with circadian rhythm disorders circadian rhythms. and insomnia. In older adults (including older nursing home residents), in Methods: Based upon a systematic grading of evidence, members of the whom traditional sleep monitoring can be difficult, actigraphy is indicated Standards of Practice Committee, including those with expertise in the for characterizing sleep and circadian patterns and to document treatment use of actigraphy, developed these practice parameters as a guide to the responses. Similarly, in normal infants and children, as well as special appropriate use of actigraphy, both as a diagnostic tool in the evaluation pediatric populations, actigraphy has proven useful for delineating sleep of sleep disorders and as an outcome measure of treatment efficacy in patterns and documenting treatment responses. clinical settings with appropriate patient populations. Conclusions: Recent research utilizing actigraphy in the assessment Recommendations: Actigraphy provides an acceptably accurate es- and management of sleep disorders has allowed the development of timate of sleep patterns in normal, healthy adult populations and inpa- evidence-based recommendations for the use of actigraphy in the clinical tients suspected of certain sleep disorders. More specifically, actigraphy setting. Additional research is warranted to further refine and broaden its is indicated to assist in the evaluation of patients with advanced sleep clinical value. phase syndrome (ASPS), delayed sleep phase syndrome (DSPS), and Keywords: Circadian rhythms, actigraphy, advanced sleep phase syn- shift work disorder. Additionally, there is some evidence to support the use drome, delayed sleep phase syndrome, shift work disorder of actigraphy in the evaluation of patients suspected of jet lag disorder and Citation: Practice Parameters for the Use of Actigraphy in the As- non-24hr sleep/wake syndrome (including that associated with blindness). sessment of Sleep and Sleep Disorders: An Update for 2007. SLEEP When polysomnography is not available, actigraphy is indicated to esti- 2007;30(4):519-529 mate total sleep time in patients with obstructive sleep apnea. In patients

1. INTRODUCTION circadian rhythm sleep disorders, sleep related breathing disorders, determination of response to therapy, and in the evaluation ACTIGRAPHY INVOLVES USE OF A PORTABLE DEVICE of sleep patterns among special populations. This literature, in THAT RECORDS MOVEMENT OVER EXTENDED PERI- combination with growing clinical experience with actigraphy, ODS OF TIME, AND HAS BEEN USED EXTENSIVELY IN led to the inclusion of actigraphy as a measure of sleep duration the study of sleep and circadian rhythms. Since the publication of and sleep patterns in the diagnostic criteria for several specific the last American Academy of Sleep Medicine (AASM) practice sleep disorders in the second edition of the International Clas- parameters on the use of actigraphy,1 there has been an explosion sification of Sleep Disorders.2 Actigraphy is listed as a diagnostic in the number of research articles utilizing actigraphy to estimate tool in the ICSD-2 primarily when sleep patterns must be assessed sleep and circadian rhythms. In response to this new literature, over time, making polysomnography impractical. For example, and the growing use of actigraphy in clinical sleep medicine, the the ICSD-2 diagnostic criteria for most circadian rhythm disor- AASM Standards of Practice Committee (SPC) undertook the de- ders requires demonstration of abnormalities in the timing of the velopment of these revised guidelines on the clinical use of this habitual sleep pattern using either actigraphy or sleep logs for technology. seven days or more. The ICSD-2 also suggests that actigraphy Since the last review, additional literature has been published may be used to document inconsistencies between objective and that addresses the use of actigraphy in the evaluation of insomnia, subjective measures of sleep timing in paradoxical insomnia, and as an aid in assessment of habitual sleep time and circadian pat- Submitted for publication January 2007 tern in patients with behaviorally induced insufficient sleep syn- Accepted for publication January 2007 drome and idiopathic hypersomnia with and without long sleep Address correspondence to: Standards of Practice Committee, American times. Actigraphy is additionally recommended as an adjunct to Academy of Sleep Medicine, One Westbrook Corporate Center, Suite 920, the Multiple Sleep Latency Test to document a stable sleep pat- Westchester IL 60154, Phone: (708) 492-0930, Fax: (780) 492-0943, Email: tern and adequate sleep times prior to the test. [email protected] However, it should be noted that although the ICSD-2 reflects consensus among experts regarding disease classification and di- SLEEP, Vol. 30, No. 4, 2007 519 Practice Parameters for the Use of Actigraphy—Morgenthaler et al agnostic criteria, there is great variability in the evidence support- the researchers know whether the intervention is active drug or ing these diagnoses and criteria. In some cases, only face validity placebo. In the case of the evaluation of actigraphy in comparison and clinical experience guided the criteria, while in others there to a reference standard (such as polysomnography), this could was a wealth of supportive research evidence. The purpose of the be interpreted as requiring that the person scoring actigraphy is present document is to provide an updated, evidence-based re- unaware of the results of polysomnography scoring. Few studies view of the use and indications for actigraphy in the evaluation actually specified whether this was the case. Given the technol- of sleep and sleep disorders. The title of the 2002 actigraphy pa- ogy used and the typical methodology currently used for scoring rameter paper was modified from that of the first one, published actigraphic recordings (computer executed scoring programs), it in 1995, which was titled: “Practice Parameters for the Use of is unlikely that researchers remembered the results of the poly- Actigraphy in The Clinical Assessment Of Sleep Disorders.” The somnograms or simultaneously reviewed both recordings. How- 2002 paper was titled: “Practice Parameters for the Role of Actig- ever, even when using computer scoring of actigraphic data, most raphy in the Study of Sleep and Circadian Rhythms: An Update situations require manual input of start and stop times. Thus, after for 2002.” This change implied an emphasis on the uses of actig- considerable discussion, the SPC elected the more conservative raphy in research. However, the current parameter paper returns approach and required an explicit declaration of blinding for a to the original focus on an evidence-based review of the use of study to receive a Level 1 rating. Some members of the Com- actigraphy in the assessment and management of sleep disorders mittee felt that this may have underestimated the quality of the in the clinical setting. evidence for use of actigraphy. 2. Reference standard. The majority of the studies evaluated 2. METHODS actigraphy in comparison to a reference standard. In some cases these were objective measures, such as polysomnography or dim The SPC of the AASM commissioned among its members light melatonin onset; in other cases the reference was subjec- those individuals with expertise in the use of actigraphy to con- tive, including sleep logs and estimates of sleep quality. For the duct this review. These content experts were appointed in January purposes of this review, we chose appropriate reference standards 2006 to review and grade evidence in the peer-reviewed scientific based on specific diagnostic categories. Reference standards for literature regarding the use of actigraphy in sleep and circadian insomnia included PSG and/or sleep logs; for circadian rhythm rhythm disorders. A computerized search was performed using sleep disorders, PSG, phase markers, and/or sleep logs; for sleep the search terms actigraph, actigraphy, actigraphic monitoring, apnea, PSG; for restless legs syndrome and periodic limb move- actigraphic recording, actimeter, actometer, wrist activity, rest ments during sleep, PSG; for infants, caregiver reported observa- activity, or sleep-wake and found 3641 titles. These were then tions; for elderly or demented persons, phase markers, sleep logs, cross-checked with 32,211 titles found using the search terms: and/or caregiver reports; and for healthy controls, PSG, phase sleep disorders, circadian rhythm, or sleep, to yield 1884 titles. markers, or sleep logs. The inclusion of research using subjective This total was then limited to those published between 2001 and reference standards (such as sleep logs, self-reported sleep, and 2005 with a minimum of 8 subjects studied by actigraphy, those in caregiver report) reflects the fact that many studies required the English, those from the core clinical journals, and those with em- study of patients over multiple sleep cycles or other circumstanc- phasis on diagnosis (using the Ovid search engine) as a modifier es where traditional PSG as a reference standard was impracti- to yield 155 articles. After review of abstracts from these articles cal (e.g., infants and nursing home residents). As such, research to determine if they met inclusion criteria, plus of articles identi- which compares actigraphy to subjective reference standards fied by pearling, a total 108 articles (see accompanying evidence does not necessarily imply a greater accuracy with either method, table) were included. Initial data extraction, preliminary evidence but it does provide evidence as to the level of agreement between grading in accordance with the standards in Table 1, and initial these methods. In addition, some studies did not compare actig- data entry into evidence tables was performed by professionals raphy with a reference standard but were useful for this review commissioned by the AASM SPC to expedite the review process. for other reasons. For example, some studies used actigraphy to This classification of evidence, based on suggestions of Sackett,3 assess treatment effects, or compared results from one actigraphy is similar to that of the prior review and practice parameter paper scoring algorithm against another. As in the prior 2002 actigra- commissioned by the AASM SPC.1 Some modifications of evi- phy review, the SPC elected to include evidence from these stud- dence level criteria were applied by the AASM SPC to this update ies which did not compare actigraphy to a reference standard but of the practice parameters for actigraphy to insure the evidence otherwise provided important information for the current review. classification was in keeping with recent updates in the literature However, there was a change in the grading criteria for these stud- for the field of evidence grading4 (see Table 1). All evidence table ies, where those studies which did not directly compare results of entries were reviewed and, if appropriate, revised by AASM SPC actigraphy with a reference standard within participants, but did content experts. Thus, all evidence grading was performed by in- provide data that allowed comparison of group means from actig- dependent review of the article by two experts, including mem- raphy data and appropriate reference standards, could be scored bers of the SPC; areas of disagreement were addressed, and, if as Level 3, rather than Level 4 or 5, as in the 2002 actigraphy needed, the chair of the AASM SPC arbitrated the final decision review (see Table 1). on evidence level. 3. What does actigraphy measure? Many studies used actig- Three methodological issues engendered considerable debate raphy data to estimate polysomnographic measures such as total and discussion by the SPC: sleep time or wake after sleep onset. However, actigraphy simply 1. Blinding. Typically, evidence graded as Level 1 according to measures movement of a limb. Although it can be highly sensitive Sackett criteria requires blinding. In evaluations of the therapeutic and there are sophisticated algorithms that purport to accurately efficacy of medications, this means that neither the patients nor estimate other parameters, it does not measure the same param- SLEEP, Vol. 30, No. 4, 2007 520 Practice Parameters for the Use of Actigraphy—Morgenthaler et al eters as an electroencephalogram. It therefore does not measure are the same as, similar to, or an expansion of recommendations sleep as it is commonly defined5 and does not measure the subjec- in the prior practice parameters are noted in the text. tive experience of sleep (as do sleep logs and questionnaires). In addition, systematic discrepancies between actigraphy and these 3. RESULTS AND RECOMMENDATIONS measures have been documented. For example, actigraphy generally underestimates sleep onset latency because many subjects are Of the 108 studies reviewed for this project (see evidence ta- inactive and awake for a period of time prior to electroencepha- ble), 44 used sleep logs alone as a reference standard, 16 used lographically defined sleep.6,7 Likewise, a recent epidemiologic polysomnography alone, and 10 used both sleep logs and poly- study reports systematic overestimation of sleep time by sleep somnography with which actigraphic ratings could be objectively logs as compared to actigraphy.8 On the other hand, insomnia pa- compared. Thirty-eight studies did not compare actigraphy to a tients frequently underestimate TST in their sleep logs.9 Reflect- reference standard, as defined in Table 1. Of the 70 studies that ing these issues, in the current report, actigraphy will generally be did compare actigraphy to a reference standard, 17 investigated described as measuring “sleep pattern” (defined as the circadian patients with circadian rhythm sleep disorders, 15 studied patients pattern of sleep and wakefulness over multiple sleep cycles) and with insomnia (including two studies of depressed patients), 11 the presence or absence of increased wake time after sleep onset. were studies of pediatric patients, 7 were studies of elderly sub- The exception to this will be in the section where total sleep time jects with and without dementia, 7 studied normal subjects, and during one night of testing is estimated by actigraphy as an aid in 5 studied patients with sleep related breathing disorders. Eight of the evaluation of sleep apnea and in the calculation of the apnea- the 70 studies were based on a variety of other patient popula- hypopnea index in patients with suspected sleep apnea. tions including 2 with nocturnal eating disorders, 2 with restless On the basis of this review the AASM SPC developed the legs syndrome, and one each of the following patient populations: recommendations included in this paper. In all but one condi- alcoholics, atypical sexual behavior during sleep, cystic fibrosis, tion, that regarding the use of actigraphy in hypersomnia, the and mixed hypersomnias. recommendations were based on evidence from studies pub- The following are recommendations of the AASM SPC and lished in peer-reviewed journals that were evaluated as noted BOD regarding the use of actigraphy in clinical practice. The re- above and specified in the description accompanying each rec- viewed literature involved a variety of actigraphic monitors and ommendation. In developing the recommendation regarding use scoring algorithms. When described in the article, the particular of actigraphy in hypersomnia, there was insufficient scientific actigraphic device and/or algorithm used are listed in the evi- data, but the SPC felt clinical guidance was indicated for use of dence tables. Clinicians using actigraphy in practice should en- actigraphy in this condition, so the Rand/UCLA Appropriate- sure that they are familiar with the operational characteristics of ness Method was used to develop the recommendation by iden- their equipment for the specific task employed. tifying the degree of agreement among the sleep experts in the SPC after review of the limited data available. The Rand/UCLA Appropriateness Method10 combines the best available scientific 3.1 Use of actigraphy in the evaluation of sleep disorders evidence with the collective judgment of experts to yield statements regarding the appropriateness of performing procedures. 3.1.1 Actigraphy is a valid way to assist in determining sleep pat- Our expert panel rated the appropriateness of this indication in terns in normal, healthy adult populations (Standard), and in pa- two rounds by individually completing rating sheets. Based on tients suspected of certain sleep disorders. (Option-Guideline- these ratings, we classified the indication as appropriate, uncer- Standard; see specific parameter below) tain, or inappropriate. We determined that if there were strict agreement that the procedure was appropriate, it would be as- This is an expansion of the previous standard (that was limited signed an “option” level recommendation. The certainty of all to the validity and reliability in detecting sleep in normal, healthy the other recommendations was assigned according to available adult populations) to include specific patient populations, such as evidence levels, as noted in Table 2. patients with insomnia and those suspected of having circadian These practice parameters define principles of practice that rhythm sleep disorders. Specific indications for actigraphy will should meet the needs of most patients in most situations. These be addressed in the parameters below. In the current review, ad- guidelines should not, however, be considered inclusive of all ditional evidence was identified supporting use of actigraphy in proper methods of care or exclusive of other methods of care rea- normal, healthy controls, and in patients with various sleep dis- sonably expected to obtain the same results. The ultimate judg- orders. Supportive studies includes nine with evidence Level 1; ment regarding appropriateness of any specific therapy must be ten with Level 2; thirty-eight Level 3; six with Level 4 and six made by the physician and patient, in light of the individual cir- graded as Level 5. The conclusion in the preponderance of studies cumstances presented by the patient, available diagnostic tools, was that actigraphy was correlated with the reference standard (as accessible treatment options, resources available, and other rel- defined in Table 1), especially for those studies rated by the SPC evant factors. at higher evidence levels. Pearson r values were reported for total The AASM expects these guidelines to have an impact on pro- sleep time comparisons between actigraphy and polysomnogra- fessional behavior, patient outcomes, and, possibly, health care phy in eight studies.7,9,11-17 The range was 0.15 to 0.92, with an costs. These practice parameters reflect the state of knowledge at simple average of 0.71. All but the lowest r values were statisti- the time of publication and will be reviewed, updated, and revised cally significant. The lowest value was reported studying patients as new information becomes available. This practice parameter suspected of sleep apnea.16 Three additional studies reported per- paper is referenced, where appropriate, with articles to support the centage agreement for total sleep time between actigraphy and recommendation(s). New recommendations, as well as those that polysomnography of 90% in normal subjects,18 84% in patients SLEEP, Vol. 30, No. 4, 2007 521 Practice Parameters for the Use of Actigraphy—Morgenthaler et al Table 1—Evidence Levels Table 2—AASM Levels of Recommendations 1. Blind, prospective comparison of results obtained by actigra- Term Definition phy to those obtained by a reference standard* on an appropri- Standard This is a generally accepted patient-care strategy, which ate spectrum of subjects and number of patients. reflects a high degree of clinical certainty. The term 2. Comparison of results obtained by actigraphy to those obtained standard generally implies the use of Level 1 evidence, by a reference standard* but blinding not specified, not pro- which directly addresses the clinical issue, or over- spective, or on a limited spectrum of subjects or number of pa- whelming Level 2 evidence. tients. Guideline This is a patient-care strategy, which reflects a moderate 3. Comparison of results obtained by actigraphy to the mean value degree of clinical certainty. The term guideline implies of a reference standard*, but not direct within-subject compari- the use of Level 2 evidence or a consensus of Level 3 son, or otherwise methodologically limited. evidence. 4. Actigraphy compared to nonstandard reference or group differ- Option This is a patient-care strategy, which reflects uncertain ences shown: clinical use. The term option implies either inconclusive a. Adequate comparison of results obtained by actigraphy to or conflicting evidence or conflicting expert opinion. those obtained by a non-standard reference*; or b. Actigraphy not compared to any reference, but actigraphy The AASM Board of Directors (BOD) approved these recommenda- results demonstrated ability to detect significant difference tions. All members of the AASM SPC and BOD completed detailed between groups or conditions in well-designed trial. conflict-of-interest statements and were found to have no conflicts of 5. Actigraphy not adequately compared to any reference, and ei- interest with regard to this subject. ther a. Actigraphy not used in a well-designed trial, or level was 3–5 in most studies included in the current review, there b. Actigraphy used in such a trial but did not demonstrate ability to detect significant difference between groups or condi- was good agreement among studies that actigraphy data correlate tions. with polysomnography (when used), sleep logs, and markers of * Reference standards for actigraphic evaluation of sleep and cir- circadian phase in patients with circadian rhythm sleep disorders. cadian rhythms varied by diagnostic category, and included gener- There were two Level 3 studies of ASPS or DSPS patients.20,21 ally accepted “gold standards,” applied in an acceptable manner. There were four studies of shift work; three were Level 3,22-24 and By diagnostic category, reference standards for insomnia included one was Level 4b.25 There was one Level 3 study of blind sub- PSG and/or sleep logs; for circadian rhythm sleep disorders, PSG, jects.26 There was one Level 327 and one Level 4b25 study of jet phase markers, and/or sleep logs; for sleep apnea, PSG; for restless lag. Finally, there was one Level 4b study of patients with non- legs syndrome and periodic limb movements during sleep, PSG; for 24-hr sleep/wake rhythm.28 infants, caregiver reported observations; for elderly or demented persons, phase markers, sleep logs, and/or caregiver reports; and for healthy controls, PSG, phase markers, or sleep logs. Nonstandard 3.1.3 When polysomnography is not available, actigraphy is indi- references include such items applied outside their diagnostic cat- cated as a method to estimate total sleep time in patients with ob- egory, or other experimental monitors. structive sleep apnea syndrome. Combined with a validated way of monitoring respiratory events, use of actigraphy may improve with sleep related breathing disorders,6 and 84% in infants.19 accuracy in assessing the severity of obstructive sleep apnea com- With the exception of the study by Penzel et al,16 most authors pared with using time in bed. (Standard) concluded that actigraphy is significantly correlated with polysomnography in the measurement of total sleep time. For exam- This parameter is a modification of the previous parameter ple, Vallieres and Morin9 concluded, “these results suggest that regarding use of actigraphy in evaluation of sleep disordered actigraphy is a reliable method for assessing sleep-wake patterns breathing, and is based on three Level 1 studies.6,14,29 Since the and for monitoring treatment response among insomnia patients.” last parameter paper, several additional studies have evaluated In a study of normal subjects, de Souza et al12 reported that “ap- both general purpose actigraphs and specially optimized actig- plying automatic sleep scoring to motor activity resulted in a raphy in patients with sleep disordered breathing. Many of the good accuracy (91%) with both the algorithms … in comparison studies have focused on the accuracy or usefulness of actigraphy to PSG.” In general, the agreement between actigraphy and poly- in estimating total sleep time (TST) in patients with sleep apnea somnography was higher than the agreement between actigraphy and combining this with tests of respiratory function in order to and sleep logs. calculate the most common measure of apnea severity, the apnea- hypopnea index (AHI). Actigraphy can provide an assessment of 3.1.2 Actigraphy is indicated to assist in the evaluation of patients TST (as it does in some other disorders), and when used along suspected of advanced sleep phase syndrome (ASPS), delayed with a valid test for the presence and type of breathing abnor- sleep phase syndrome (DSPS), and shift work sleep disorder mality, can improve the calculation of AHI compared with us- (Guideline); and circadian rhythm disorders, including jet lag and ing time in bed. Several other studies used actigraphy as part of non-24-hour sleep/wake syndrome [including that associated with research protocols evaluating sleep pattern of patients with OSA blindness] (Option) without actually comparing actigraphy results to a sleep standard. No studies propose actigraphy alone as a method of determine the This is a modification of the recommendation from the prior presence of sleep apnea. practice parameter paper and expands the role of actigraphy in the One study (Level 1)14 found a high correlation (r = 0.90, P = diagnosis of circadian rhythm sleep disorders. The use of actigra- 0.0001) between TST measured by PSG (pTST) and TST esti- phy for evaluation of circadian rhythm disorders is based on ad- mated by actigraphy (aTST) in patients with obstructive sleep ditional evidence included in this review. Although the evidence apnea syndrome. Agreement using the Bland and Altman method SLEEP, Vol. 30, No. 4, 2007 522 Practice Parameters for the Use of Actigraphy—Morgenthaler et al found the difference between pTST and aTST was only 2.5 min, the study was otherwise methodologically limited for estimating but there were notable overestimations and underestimations in the utility of actigraphy. three of the 26 patients. In another study, Elbaz et al29 (Level 1) also found excellent correlation between pTST and aTST (r = 3.1.4. Actigraphy is indicated as a method to characterize circadian 0.74, P <0.0001). In the latter study, the AHI, calculated as the ap- rhythm patterns or sleep disturbances in individuals with insomnia, neas plus hypopneas per hour of actigraphically determined sleep including insomnia associated with depression. (Option) (aAHI) was compared with PSG results, again showing excellent correlation (r = 0.976, P <0.0001). The aAHI was more accurate This is similar to the prior recommendation. There were two than an AHI determined by dividing the apneas plus hypopneas by Level 5b studies33,34 characterizing sleep patterns in individuals time in bed, indicating that the addition of actigraphy improved with complaints of insomnia. There were two Level 335,36 studies accuracy when estimating the AHI without EEG measured sleep and one Level 4b37 study indicating that actigraphy is a way to time. In both of these studies, the accuracy of actigraphy evalu- characterize sleep or circadian rhythms in patients with a depres- ated using Bland Altman methods declined in patients with more sive disorder. severe sleep apnea, but in the study of Elbaz et al,29 only 1 of 20 patients were overclassified with respect to OSA severity by 3.1.5 Actigraphy is indicated as a way to determine circadian pat- the aAHI (severe instead of moderate severity), and none were tern and estimate average daily sleep time in individuals complain- underclassified. Thus, it appears that even though the estimate of ing of hypersomnia (Option). TST becomes less accurate as apneas and hypopneas increase, the actigraphically derived AHI in most cases accurately classifies There were no studies identified that compared actigraphy ver- moderate or severe sleep apnea. Actigraphically estimated TST in sus the clinical history plus sleep logs (or another reference stan- milder cases of sleep apnea appear to be quite accurate, especially dard) to estimate mean sleep time or sleep pattern when evaluating if using specially optimized actigraphs and evaluation algorithms. patients with hypersomnia as a complaint. One Level 3 study eval- These Level 1 studies were performed in a sleep laboratory, and uated patients diagnosed with a variety of hypersomnia disorders, extrapolation to the home environment could introduce issues not including narcolepsy, idiopathic hypersomnia, hypersomnia asso- anticipated. However, most other studies reviewed for this paper ciated with psychiatric disorders, HIV-encephalopathy, brainstem involved use of actigraphy outside the sleep laboratory and had stroke, periodic hypersomnia, postviral illness, and head trauma.38 low data failure rates. Because of the complexity of data analysis, Actigraphy was used to determine the average daily sleep time over evaluations of sleep disordered breathing severity that use actig- one week prior to evaluation with PSG and MSLT, and biochemical raphy to estimate TST should be interpreted with caution by ex- assessment. Actigraphy estimated mean sleep time varied between perienced sleep clinicians who are familiar with the performance diagnostic groups, with patients with hypersomnia associated with characteristics of the particular actigraphic system employed. psychiatric disorders sleeping longer on average (P <0.037 by Wil- Another Level 1 study evaluated 228 patients using a special coxon rank sums method, our own analysis of their data). Acti- actigraphic system optimized to patients with suspected sleep dis- graphically determined TST averaged ≥9 hours per day in 11 of 27 ordered breathing.6 Using epoch by epoch comparison of sleep patients, including all but one patient with hypersomnia associated versus wake determined actigraphically versus PSG across all with psychiatric disorders, and none of the patients diagnosed with subjects, sensitivity of detecting sleep was 88.8%, specificity narcolepsy or idiopathic hypersomnia. The shortest mean aTST was 69.5%, and agreement was 84%. Sensitivity and agreement was 7.44 hours per day in the idiopathic hypersomnia group. The tended to go down with increasing SDB levels (from 91% to 85%, authors indicated that history plus the results of actigraphy, PSG, and 86% to 79%, respectively). Specificity was less affected by and MSLT contributed to the diagnosis of disorders of hypersom- increasing SDB levels (ranged between 68% and 71%). Consider- nolence, but the exact role of actigraphy in interpreting MSLT or ing all subjects, aTST versus pTST was 690 ± 152 and 690 ± 154 assigning diagnoses was not described. minutes, respectively (P >0.05). However, a Level 2 study16 utiliz- The complaint of sleepiness must be evaluated in the context of ing the same optimized device found no significant correlation be- recent sleep duration and pattern before a judgment can be made tween pTST and aTST. Bland-Altman comparison showed much as to the pathologic nature of the complaint. The guidelines de- scatter, with mean of the differences in TST = 12.17 + 64.5 min. veloped for the MSLT39 indicate that sleep logs may be obtained Another (Level 4b) study30 using the same device found a good for 1 week prior to the PSG/MSLT to assess sleep-wake sched- correlation between the arousal index estimated from a device us- ules and assist in interpretation of results, while the ICSD-2 indi- ing peripheral arterial tonometry changes to detect arousal and ac- cates that “the sleep-wake schedule must have been standardized tigraphy to estimate sleep time, and the arousal index determined for at least seven days before the polysomnographic testing (and by conventional PSG methods (r = 0.87, P <0.0001). In this study documented by sleep log or actigraphy)” in order to properly in- there was no report of actual TST or number of PAT arousals; only terpret an MSLT. However, some individuals, such as those with the ratio was reported. Therefore, the contribution of actigraphy impaired cognition, literacy, or motivation may be unable to keep to the reported correlation could not be evaluated. accurate sleep logs, and both over- and underreporting of total Finally, two studies used actigraphy to estimate TST in patients sleep time and pattern have been of concern. Therefore, the com- with sleep disordered breathing without formal comparison to an- mittee used the Rand/UCLA Appropriateness Method (described other measure of TST. Larkin et al (Level 4b)31 found that mean above) to determine expert consensus regarding this parameter TST correlated with changes in C-reactive protein in adolescents on the indications for use of actigraphy in hypersomnia. There with sleep apnea. Noseda et al (Level 5a)32 used leg actigraphy to was agreement that actigraphy is an appropriate way to ensure measure treatment-induced changes in leg activity in patients with stable sleep patterns and adequate sleep duration prior to PSG sleep disordered breathing, and did detect treatment effects, but and MSLT. SLEEP, Vol. 30, No. 4, 2007 523 Practice Parameters for the Use of Actigraphy—Morgenthaler et al 3.2 Use of actigraphy in assessing the response to therapy of sleep older adults living in the community is based on two additional disorders Level 3 studies.51,52 In a placebo-controlled trial51 of melatonin treatment in healthy older adults presenting either with or without 3.2.1 Actigraphy is useful as an outcome measure in evaluating the sleep complaints, sleep diaries were used as the reference stan- response to treatment for circadian rhythm disorders. (Guideline) dard. There was little difference in subjective and actigraphically estimated sleep quality on either measure in either group as a re- This is the same as the recommendation in the previous prac- sult of melatonin treatment. In the other study,52 both in-labora- tice parameters paper. Additional evidence shows that changes in tory and at-home measures were taken to determine the effects of actigraphy measures are in agreement with other outcome mea- daytime naps on nocturnal sleep and performance. In the at-home sures in the assessment of response to intervention in patients condition, TST and sleep efficiency were consistent when com- with circadian rhythm sleep disorders. pared between actigraphy and sleep log results. There were two additional Level 3 studies using actigraphy as an outcome measure in the treatment of jet lag40,41 and one addi- 3.3.2 Actigraphy is indicated for characterizing and monitoring tional Level 3 study using actigraphy as an outcome measure in a sleep and circadian rhythm patterns and to document treatment study of shift work.42 outcome (in terms of sleep patterns and circadian rhythms) among older nursing home residents (in whom traditional sleep monitor- 3.2.2 Actigraphy is useful for evaluating the response to treatment ing by polysomnography can be difficult to perform and/or inter- for patients with insomnia, including insomnia associated with de- pret). (Guideline) pressive disorders. (Guideline) This is a modification of the recommendation of the previous This is the same as the recommendation from the previous practice parameter paper. The evidence for the use of actigraphy practice parameter paper. There were one additional Level 1,43 to characterize and monitor sleep and circadian rhythm patterns two Level 2,9,44 two Level 3,45,46 and two Level 5b studies,33,34 in- among older adults living in nursing homes is based on five ad- dicating that actigraphy is useful in detecting treatment response ditional studies in the nursing home setting. Two Level 3 stud- in people diagnosed with insomnia. In addition there were two ies,53,54 two Level 4 studies,55,56 and one Level 5b study57 were Level 347,48 studies indicating that actigraphy is a useful adjunct in identified. Several studies used observer ratings as the reference detecting treatment response in people diagnosed with disrupted standard for comparison with actigraphy. One Level 3 study53 sleep or circadian rhythms associated with a depressive disorder. found that although results were similar between nursing staff and actigraphy for some sleep measures, nursing staff noted less 3.3 Use of actigraphy in special populations and special situations sleep disruption during the night (WASO) than was recorded by actigraphy. Another Level 3 study54 was able to discriminate 3.3.1 Actigraphy is useful for characterizing and monitoring sleep diagnostic subtypes among dementia patients according to pat- and circadian rhythm patterns and to document treatment outcome terns of activity and core body temperature rhythms. Two Level (in terms of sleep patterns and circadian rhythms) among older 4 studies examined patterns of rest/activity in relation to presence adults living in the community, particularly when used in conjunc- or absence or Level of dementia. A Level 4b study55 found that tion with other measures such as sleep diaries and/or caregiver actigraphic rest/activity patterns differentiated patients with mild observations. (Guideline) dementia from those advanced to the moderate stage. Similarly, a Level 4a study56 was able to distinguish demented from non- This recommendation is a modification of the previous practice demented subjects on the basis of daytime and nocturnal activity parameter paper. The evidence for use of actigraphy to character- levels. Further they found that functional ability was associated ize and monitor sleep and circadian rhythm patterns among older with diurnal patterns of activity. adults living in the community is based on two additional studies The evidence for use of actigraphy to document treatment out- identified in the current review that addressed the use of actig- come (in terms of sleep patterns and circadian rhythms) among raphy in normal older adults. There were one Level 249 and one older nursing home residents is based on 13 additional treatment Level 350 study using actigraphy to evaluate sleep and circadian outcome studies, including two Level 217,58 and three Level 3 rhythms in normal older people. In the Level 2 study by Ceolim et studies.59-61 Furthermore, there were six Level 462-67 and two Level al49 there were significant correlations (P <0.005) between sleep 5b68,69 studies evaluating treatment outcomes in dementia or nurs- log and actigraphic variables (e.g., TST) collected for 23 days ing home populations. One Level 2 study58 tested the effects of in over 76% of a sample of healthy older people. In the Level 3 withdrawal of antipsychotic medication on sleep/wake activity study of a sample of 103 community-dwelling older adults,50 ac- and on behavioral and psychological symptoms in nursing home tigraphic measures correlated with subjective reports in subjects residents.58 Actigraphic results were compared with psychiatric without sleep complaints but not in those complaining about their inventory responses, and restlessness was significantly associated sleep. Although able to distinguish between noncomplaining good with mean 24-hr actigraphic measures of activity (r = 0.60, P = sleepers and complaining poor sleepers, actigraphy was not able 0.001) and nocturnal sleep problems were significantly associ- to distinguish between other categories of sleepers in this sample. ated with nighttime activity levels. (r = 0.60, P = 0.001). In an- Results of this study provided evidence of actigraphy’s ability to other Level 2 study17 of a randomized controlled trial comparing determine TST and sleep onset latency (in women only) for those the effects of two different doses of melatonin and placebo on not able to provide sleep diary information. sleep in Alzheimer disease patients found no significant differ- The evidence for use of actigraphy to document treatment out- ences on sleep outcome by actigraphy between treatment groups. come (in terms of sleep patterns and circadian rhythms) among However, a subset of seven subjects had simultaneous actigraphy SLEEP, Vol. 30, No. 4, 2007 524 Practice Parameters for the Use of Actigraphy—Morgenthaler et al and PSG for a period of 18 days and the TST estimated by actig- tween light exposure, sleep patterns, and crying in healthy 6-12 raphy correlated highly with PSG ( r= 0.92, P <0.01). In addition, week old infants found overall consistency between actigraphic a Level 3 study59 testing the effects of bright light in a nursing measures of nocturnal activity and parental reports of sleep. A home sample found significant improvements in sleep time and third Level 3 study documented some differences in activity-rest wake time within nocturnal sleep according to actigraphy which cycles but not in other sleep parameters during the first week of paralleled nursing staff ratings. life in infants grouped according to delivery mode (planned C- section, emergency Caesarian section, and normal spontaneous 3.3.3 Actigraphy is indicated for delineating sleep patterns, and to vaginal delivery).74 One Level 4b study75 found some significant document treatment responses in normal infants and children (in differences (i.e., increased variability ultradian cycles, diurnal whom traditional sleep monitoring by polysomnography can be dif- sleep duration) in actigraphically derived activity-rest behaviors ficult to perform and/or interpret), and in special pediatric popula- between healthy pre-term and full-term infants, while another76 tions. (Guideline) Level 4b study used actigraphy to characterize inter-individual variability in activity-rest behavior and differences in sleep du- This recommendation is a modification of the recommendation ration between pre- and full-term infants. Actigraphy was also from the previous practice parameter paper.1 This recommenda- used in a Level 4b study77 to document a significant increase in tion is based on 23 additional studies identified in the current re- nocturnal activity counts associated with rapid ascent to moderate view that addressed the use of actigraphy in children. There were altitudes in a groups of infants and young children (4-33 months), a total of five Level 2 studies (no studies were identified as Level and in another Level 4b study78 which examined the development 1, due to the absence of information regarding blinding, as de- of circadian rhythms in newborns by comparing sleep-wake pat- scribed above), seven Level 3 studies, nine Level 4 studies, and terns longitudinally in newborns and their mothers at 3, 6, 9, and two Level 5 studies of actigraphy in pediatric populations. These 12 weeks. A Level 5 study19 used actigraphy to determine sleep- studies included a range of age groups (infant through adoles- ing position and measure sleep-wake patterns in healthy 34-42 cent), as well as a number of different medical, psychiatric, and week old infants. Finally, a Level 3 study79 assessing differences sleep disordered diagnostic groups, and used a variety of refer- in sleep patterns in parents of newborns, found that mothers had ence standards. less actigraphically documented sleep at night and more during In terms of age groups, the largest numbers of studies (10) the day compared to fathers, that breastfeeding was associated were focused on infants (typically between 6 and 12 months). One with more WASO, and that working mothers had an average 6-7 Level 2 study70 compared a parent-report infant sleep question- minutes less sleep in 24 hours than nonworking mothers. naire (Brief Infant Sleep Questionnaire – BISQ) with actigraphy Older children and adolescents were subjects in several other and daily sleep logs to assess correspondence between measures, studies. A Level 4b study80 which assessed the ability of measures as well as to determine differences between a control and clini- of emotional intensity (maternal rating, vagal functioning) to pre- cal sample of infants referred to a sleep clinic. Significant but dict actigraphically determined sleep problems in healthy school- moderate correlations were found between BISQ and actigraphic aged children, found that increased emotional intensity was cor- measures of sleep onset latency (SOL) (r = 0.54, P <0.001) and related with reduced nocturnal sleep and increased night activity. night wakings (r = 0.42, P <0.0001), with nocturnal sleep dura- A Level 2 study81 examined the validity of a self-report adolescent tion showing lower agreement (r = 0.23, P <0.05). In contrast, the sleep survey by comparing retrospective self-report estimates of most robust correlations found between actigraphy measures and sleep patterns (TST, bedtime, and waketime on weekends and the reference standard (daily sleep logs) were found for SOL (r = weekdays) with sleep parameters measured by both actigraphy 0.96, P <0.0001) and nocturnal TST (r = 0.87, P <0.0001), rather and sleep logs over a subsequent week. Survey-estimated school- than night wakings (r = 0.49, P <0.0001). There were also some night total sleep times and wake times did not differ from diary significant systematic differences between actigraphic and sleep and actigraphy measures, although survey bedtimes were slightly log measures, with actigraphy providing lower estimates of sleep earlier. On weekends, survey-reported sleep duration was about duration and higher estimates of night wakings compared to sleep 30 minutes longer than estimated by sleep diaries (t = 4.26, P diaries. Only one actigraphic measure, number of night wakings, <0.001) and actigraphy (t = 5.25, P <0.001), and wake times were had a unique contribution in discriminating between the control about 55 minutes longer. Overall, school- and weekend-night sur- and clinical samples (F=6.29, P <0.05). vey variables were significantly correlated with both diary and ac- A different reference standard, direct observation of infant be- tigraphy variables, but the strength of the associations were con- havioral states, was used in a Level 2 study using actigraphy in as- sistently greater for school-night variables than for corresponding sessing sleep-wake rhythm and sleep structure in healthy 1, 3, and weekend-night variables. However, it should be noted that there 6 month old infants.71 The overall agreement between measures was no attempt to directly compare actigraphy and the reference in scoring sleep and wake was satisfactory (between 87% and standard sleep log variables in this study; in fact, it was noted in 95%) after 3 months of age, but agreement was less than 73% at 1 the Methods section that the procedure (“Sadeh algorithm”) used month. Reliable actigraphic distinction, however, between active to analyze actigraphy “relies heavily upon the concurrent behav- and quiet sleep could not be made in any of the three age groups. ioral self-report obtained by the sleep diaries,” and thus the two Healthy term 6-8 week old infants were also the subjects in measures would be expected to be highly correlated. a Level 3 study72 which assessed the effects of infant massage Actigraphic measures of sleep were also used in studies of sev- on the development of circadian rhythms by comparing actigra- eral pediatric patient populations with chronic medical conditions. phy and salivary melatonin levels; peaks of period activity were One was a Level 2 study that primarily assessed the relationship delayed in the intervention group compared to controls. Another between sleep disturbance and pulmonary function in a group of Level 3 study73 which longitudinally assessed the relationship be- children with cystic fibrosis (CF) but also compared actigraphy SLEEP, Vol. 30, No. 4, 2007 525 Practice Parameters for the Use of Actigraphy—Morgenthaler et al to parent- and self-report data in this population.82 There was a 4. RECOMMENDATIONS FOR FUTURE RESEARCH significant correlation between sleep duration (the only parameter reported) as measured by actigraphy with sleep period reported by 4.1 parents (r = 0.79, P<0.0001) and by children (r = 0.71, P <0.0001) in the control group, but not in the CF group (r = 0.29, P = 0.06; Additional research is needed which compares results from r = 0.18, P = 0.2, respectively). A Level 3 study83 used actigraphy different actigraphy devices and the variety of algorithms used to confirm sleep instability, frequent microarousals, and increased to evaluate actigraphy data in order to further establish standards daytime napping in a group of children with Smith-Magenis syn- of actigraphy technology. Well designed studies using actigraphy drome (a genetic syndrome frequently characterized by self-in- should describe the device and the analysis algorithms used. jury and sleep disturbances). Actigraphy was also used to measure sleep disturbance in a Level 4b study84 of blind adolescents with 4.2 and without optic nerve disease, which documented that greater wake time instability was associated with optic nerve disease. There is need for additional study addressing the reliability and In studies of children with psychiatric disorders, one Level 3 validity of actigraphy compared to reference standards, such as study85 used actigraphy to study sleep patterns in children with polysomnography, and the circadian rhythms of basic physiologic ADHD with and without sleep problems compared to controls, functions, such as temperature, cortisol, and melatonin levels. and found significantly delayed sleep onset and offset in children with ADHD and insomnia, suggesting a circadian rhythm abnor- 4.3 mality. A Level 4b86 intervention study used actigraphy to document treatment response (decrease in mean nocturnal activity) to Further research is needed to establish standards for setting melatonin and rebound sleep disturbance following discontinua- start and stop times of the sleep and wake periods when using tion in children with Asperger syndrome. Another Level 4b35 study, actigraphy, including techniques such as event markers or sleep which used actigraphy to evaluate locomotor activity and circadian diaries, and other methods in the study of populations where these rest-activity cycles in children with major depression compared to techniques may not be valid (e.g., dementia patients, nursing controls, found significant differences related to gender and age but home setting). For example, difficulty in establishing a standard not group assignment. Finally, a retrospective chart review (Level for setting start time is likely one factor contributing to the diffi- 5 study)87 of children with ADHD referred to a sleep center showed culty in correlating certain sleep variables (especially sleep onset a high incidence (94%) of sleep onset delay and high night-to-night latency) measured by actigraphy with findings from PSG. variability in sleep patterns in the small percentage (16%) of subjects for whom actigraphy data was available. 4.4 There were also several studies which used actigraphy to assess sleep in children with sleep disorders. Agreement between Well-designed studies should include technical details related periodic limb movement during sleep scored by actigraphy and to the administration and scoring of actigraphy. In much of the ex- those detected with anterior tibialis EMG was assessed in a Level isting literature, there is an inadequate description of whether vi- 2 study of ninety-nine 4- to 12-year-old children.88 It was con- sual inspection of data is performed, how missing data is handled, cluded that this actigraphic measurement of PLMs in children was and other important decisions made in the analysis of actigraphy not sufficiently accurate to permit use in clinical settings. Spe- data. More research is needed to assess the reliability of actigra- cifically, actigraphy tended to overestimate PLMs compared to phy under various clinical circumstances, and to determine what EMG, and, although the application of a correction factor based parameters may be used to assess the quality of actigraphic data. on average number of EMG-derived movement during arousals improved agreement between measures somewhat, different cor- 4.5 rection factors were required for each of the different diagnostic groups (SDB, primary snoring/normal, and periodic limb move- Further work is needed to clarify the relative and unique con- ment disorder), limiting its utility as a diagnostic measure. One tributions of actigraphy, polysomnography and sleep logs in the Level 4b study31 of adolescents with SDB found sleep duration diagnosis of sleep disorders and measurement of treatment ef- was significantly negatively correlated with C-reactive protein, fects. For example, besides estimates of wake and sleep times, body mass index, and AHI. there are various other data generated by commercially available Finally, one Level 2 study89 compared actigraph placement analysis software, such as fragmentation index and movement in- (waist vs nondominant wrist) in estimating sleep duration in dex, for which clinical correlates are not well described. school-aged children. Although diurnal activity was lower with waist placement, the overall minute-by-minute agreement of 4.6 sleep-wake states between placement sites was 92.5% (range 82.3%–97.7%), and nocturnal agreement was 95.6%. None of the The use of actigraphy in hypersomnia populations, especially mean sleep estimates (sleep duration, sleep latency, sleep percent- as an adjunct to the Multiple Sleep Latency Test, should be tested age, sleep efficiency) were significantly affected by placement to establish an evidence-based recommendation for the use of site, although there were some inter-individual differences in actigraphy in the clinical evaluation and management of hyper- agreement (sleep duration and latency). Another Level 3 study90 somnia. assessing compliance with imposed sleep schedules in the home setting in school-aged children demonstrated significant differences in actigraphically measured sleep according to condition. SLEEP, Vol. 30, No. 4, 2007 526 Practice Parameters for the Use of Actigraphy—Morgenthaler et al sleep-wake identification in infants. Physiol Meas 2004;25:1291- ACKNOWLEDGEMENTS 304. 20. Ando K, Kripke DF, Ancoli-Israel S. Delayed and advanced sleep The AASM and the SPC would like to thank Maria DeSena phase symptoms. Isr J Psychiatry Relat Sci 2002;39:11-8. and Susan Benloucif, PhD, for coordinating the work on this 21. Kripke DF, Youngstedt SD, Elliott JA, et al. Circadian phase in practice parameter and to Richard Rosenberg, PhD, and Brian adults of contrasting ages. Chronobiol Int 2005;22:695-709. Shearer, PhD, for contributions to the preparation of the manu- 22. Borges FN, Fischer FM. Twelve-hour night shifts of healthcare script. workers: a risk to the patients? Chronobiol Int 2003;20:351-60. 23. Lamond N, Darwent D, Dawson D. How well do train driver’s sleep REFERENCES in relay vans? Ind Health 2005;43:98-104. 24. Daurat A, Foret J. Sleep strategies of 12-hour shift nurses with 1. Littner M, Kushida CA, Anderson WM, et al. Practice parameters for emphasis on night sleep episodes. Scand J Work Environ Health the role of actigraphy in the study of sleep and circadian rhythms: an 2004;30:299-305. update for 2002. Sleep 2003;26:337-41. 25. Carvalho Bos S, Waterhouse J, Edwards B, Simons R, Reilly T. The 2. American Academy of Sleep Medicine. International classification use of actimetry to assess changes to the rest-activity cycle. Chro- of sleep disorders: Diagnostic and coding manual. 2nd ed. West- nobiol Int 2003;20:1039-59. chester, IL: American Academy of Sleep Medicine; 2005. 26. Leger D, Guilleminault C, Santos C, Paillard M. Sleep/wake cycles 3. Sackett DL. Rules of evidence and clinical recommendations for the in the dark: sleep recorded by polysomnography in 26 totally blind management of patients. Can J Cardiol 1993;9:487-9. subjects compared to controls. Clin Neurophysiol 2002;113:1607- 4. Atkins D, Eccles M, Flottorp S, et al. Systems for grading the qual- 14. ity of evidence and the strength of recommendations I: critical ap- 27. Grajewski B, Nguyen MM, Whelan EA, Cole RJ, Hein MJ. Mea- praisal of existing approaches The GRADE Working Group. BMC suring and identifying large-study metrics for circadian rhythm dis- Health Serv Res 2004;4:38. ruption in female flight attendants. Scand J Work Environ Health 5. Rechtschaffen A, Kales A. A manual of standardized terminology, 2003;29:337-46. techniques and scoring system for sleep stages of human subjects. 28. Uchiyama M, Shibui K, Hayakawa T, et al. Larger phase angle be- Los Angeles, CA: Brain Information Service/Brain Research Insti- tween sleep propensity and melatonin rhythms in sighted humans tute; 1968. with non-24-hour sleep-wake syndrome. Sleep 2002;25:83-8. 6. Hedner J, Pillar G, Pittman SD, Zou D, Grote L, White DP. A novel 29. Elbaz M, Roue GM, Lofaso F, Quera Salva MA. Utility of actigra- adaptive wrist actigraphy algorithm for sleep-wake assessment in phy in the diagnosis of obstructive sleep apnea. Sleep 2002;25:527- sleep apnea patients. Sleep 2004;27:1560-6. 31. 7. Lichstein KL, Stone KC, Donaldson J, et al. Actigraphy validation 30. Pillar G, Bar A, Betito M, et al. An automatic ambulatory device for with insomnia. Sleep 2006;29:232-9. detection of AASM defined arousals from sleep: the WP100. Sleep 8. Lauderdale DS, Knutson KL, Yan LL, et al. Objectively measured Med 2003;4:207-12. sleep characteristics among early-middle-aged adults: the CARDIA 31. Larkin EK, Rosen CL, Kirchner HL, et al. Variation of C-reactive study. Am J Epidemiol 2006;164:5-16. protein levels in adolescents: association with sleep-disordered 9. Vallieres A, Morin CM. Actigraphy in the assessment of insomnia. breathing and sleep duration. Circulation 2005;111:1978-84. Sleep 2003;26:902-6. 32. Noseda A, Nouvelle M, Lanquart JR, et al. High leg motor activ- 10. Fitch K, Bernstein SJ, Aguilar MS, et al. The RAND/UCLA Ap- ity in sleep apnea hypopnea patients: efficacy of clonazepam com- propriateness Method User’s Manual. Santa Monica, CA: RAND bined with nasal CPAP on polysomnographic variables. Respir Med Corporation; 2001. 2002;96:693-9. 11. Currie SR, Malhotra S, Clark S. Agreement among subjective, ob- 33. Nelson J, Harvey AG. The differential functions of imagery and jective, and collateral measures of insomnia in postwithdrawal re- verbal thought in insomnia. J Abnorm Psychol 2002;111:665-9. covering alcoholics. Behav Sleep Med 2004;2:148-61. 34. Currie SR, Clark S, Hodgins DC, El-Guebaly N. Randomized con- 12. de Souza L, Benedito-Silva AA, Pires ML, Poyares D, Tufik S, trolled trial of brief cognitive-behavioural interventions for insom- Calil HM. Further validation of actigraphy for sleep studies. Sleep nia in recovering alcoholics. Addiction 2004;99:1121-32. 2003;26:81-5. 35. Armitage R, Hoffmann R, Emslie G, Rintelman J, Moore J, Lewis 13. Edinger JD, Means MK, Stechuchak KM, Olsen MK. A pilot K. Rest-activity cycles in childhood and adolescent depression. J study of inexpensive sleep-assessment devices. Behav Sleep Med Am Acad Child Adolesc Psychiatry 2004;43:761-9. 2004;2:41-9. 36. Harvey AG, Schmidt DA, Scarna A, Semler CN, Goodwin GM. 14. Gagnadoux F, Nguyen XL, Rakotonanahary D, Vidal S, Fleury B. Sleep-related functioning in euthymic patients with bipolar disor- Wrist-actigraphic estimation of sleep time under nCPAP treatment der, patients with insomnia, and subjects without sleep problems. in sleep apnoea patients. Eur Respir J 2004;23:891-5. Am J Psychiatry 2005;162:50-7. 15. Lotjonen J, Korhonen I, Hirvonen K, Eskelinen S, Myllymaki M, 37. Korszun A, Young EA, Engleberg NC, Brucksch CB, Greden JF, Partinen M. Automatic sleep-wake and nap analysis with a new Crofford LA. Use of actigraphy for monitoring sleep and activity wrist worn online activity monitoring device vivago WristCare. levels in patients with fibromyalgia and depression. J Psychosom Sleep 2003;26:86-90. Res 2002;52:439-43. 16. Penzel T, Kesper K, Pinnow I, Becker HF, Vogelmeier C. Peripheral 38. Bassetti C, Gugger M, Bischof M, et al. The narcoleptic borderland: arterial tonometry, oximetry and actigraphy for ambulatory record- a multimodal diagnostic approach including cerebrospinal fluid lev- ing of sleep apnea. Physiol Meas 2004;25:1025-36. els of hypocretin-1 (orexin A). Sleep Med 2003;4:7-12. 17. Singer C, Tractenberg RE, Kaye J, et al. A multicenter, placebo- 39. Littner MR, Kushida C, Wise M, et al. Practice parameters for clini- controlled trial of melatonin for sleep disturbance in Alzheimer’s cal use of the multiple sleep latency test and the maintenance of disease. Sleep 2003;26:893-901. wakefulness test. Sleep 2005;28:113-21. 18. Matsumoto M, Miyagishi T, Sack RL, Hughes RJ, Blood ML, 40. Beaumont M, Batejat D, Pierard C, et al. Caffeine or melatonin Lewy AJ. Evaluation of the Actillume wrist actigraphy monitor effects on sleep and sleepiness after rapid eastward transmeridian in the detection of sleeping and waking. Psychiatry Clin Neurosci travel. J Appl Physiol 2004;96:50-8. 1998;52:160-1. 41. Boulos Z, Macchi MM, Sturchler MP, et al. Light visor treatment 19. Sazonov E, Sazonova N, Schuckers S, Neuman M. Activity-based for jet lag after westward travel across six time zones. Aviat Space SLEEP, Vol. 30, No. 4, 2007 527 Practice Parameters for the Use of Actigraphy—Morgenthaler et al Environ Med 2002;73:953-63. sephson KR. Randomized, controlled trial of a nonpharmacological 42. Hilliker NA, Muehlbach MJ, Schweitzer PK, Walsh JK. Sleepiness/ intervention to improve abnormal sleep/wake patterns in nursing alertness on a simulated night shift schedule and morningness-eve- home residents. J Am Geriatr Soc 2005;53:803-10. ningness tendency. Sleep 1992;15:430. 63. Asayama K, Yamadera H, Ito T, Suzuki H, Kudo Y, Endo S. Double 43. Wilson SJ, Rich AS, Rich NC, Potokar J, Nutt DJ. Evaluation of blind study of melatonin effects on the sleep-wake rhythm, cogni- actigraphy and automated telephoned questionnaires to assess hyp- tive and non-cognitive functions in Alzheimer type dementia. J Nip- notic effects in insomnia. Int Clin Psychopharmacol 2004;19:77- pon Med Sch 2003;70:334-41. 84. 64. Dowling GA, Hubbard EM, Mastick J, Luxenberg JS, Burr RL, Van 44. Rybarczyk B, Lopez M, Benson R, Alsten C, Stepanski E. Efficacy Someren EJ. Effect of morning bright light treatment for rest-activ- of two behavioral treatment programs for comorbid geriatric insom- ity disruption in institutionalized patients with severe Alzheimer’s nia. Psychol Aging 2002;17:288-98. disease. Int Psychogeriatr 2005;17:221-36. 45. Tang NK, Harvey AG. Correcting distorted perception of sleep 65. Fetveit A, Bjorvatn B. The effects of bright-light therapy on acti- in insomnia: a novel behavioural experiment? Behav Res Ther graphical measured sleep last for several weeks post-treatment. A 2004;42:27-39. study in a nursing home population. J Sleep Res 2004;13:153-8. 46. Semler CN, Harvey AG. Misperception of sleep can adversely af- 66. McCurry SM, Gibbons LE, Logsdon RG, Vitiello MV, Teri L. fect daytime functioning in insomnia. Behav Res Ther 2005;43:843- Nighttime insomnia treatment and education for Alzheimer’s dis- 56. ease: a randomized, controlled trial. J Am Geriatr Soc 2005;53:793- 47. Coffield TG, Tryon WW. Construct validation of actigraphic sleep 802. measures in hospitalized depressed patients. Behav Sleep Med 67. Skjerve A, Holsten F, Aarsland D, Bjorvatn B, Nygaard HA, Jo- 2004;2:24-40. hansen IM. Improvement in behavioral symptoms and advance of 48. Winkler D, Pjrek E, Praschak-Rieder N, et al. Actigraphy in patients activity acrophase after short-term bright light treatment in severe with seasonal affective disorder and healthy control subjects treated dementia. Psychiatry Clin Neurosci 2004;58:343-7. with light therapy. Biol Psychiatry 2005;58:331-6. 68. Ancoli-Israel S, Gehrman P, Martin JL, et al. Increased light expo- 49. Ceolim MF, Menna-Barreto L. Sleep/wake cycle and physical activ- sure consolidates sleep and strengthens circadian rhythms in severe ity in healthy elderly people. Sleep Res Online 2000;3:87-95. Alzheimer’s disease patients. Behav Sleep Med 2003;1:22-36. 50. McCrae CS, Rowe MA, Tierney CG, Dautovich ND, Definis AL, 69. Scherder E, Knol D, van Someren E, et al. Effects of low-frequen- McNamara JP. Sleep complaints, subjective and objective sleep pat- cy cranial electrostimulation on the rest-activity rhythm and sali- terns, health, psychological adjustment, and daytime functioning in vary cortisol in Alzheimer’s disease. Neurorehabil Neural Repair community-dwelling older adults. J Gerontol B Psychol Sci Soc Sci 2003;17:101-8. 2005;60:P182-9. 70. Sadeh A. A brief screening questionnaire for infant sleep problems: 51. Baskett JJ, Broad JB, Wood PC, et al. Does melatonin improve validation and findings for an Internet sample. Pediatrics 2004;113: sleep in older people? A randomised crossover trial. Age Ageing e570-7. 2003;32:164-70. 71. Gnidovec B, Neubauer D, Zidar J. Actigraphic assessment of sleep- 52. Monk TH, Buysse DJ, Carrier J, Billy BD, Rose LR. Effects of af- wake rhythm during the first 6 months of life. Clin Neurophysiol ternoon “siesta” naps on sleep, alertness, performance, and circa- 2002;113:1815-21. dian rhythms in the elderly. Sleep 2001;24:680-7. 72. Ferber SG, Laudon M, Kuint J, Weller A, Zisapel N. Massage 53. Fetveit A, Bjorvatn B. Sleep disturbances among nursing home resi- therapy by mothers enhances the adjustment of circadian rhythms dents. Int J Geriatr Psychiatry 2002;17:604-9. to the nocturnal period in full-term infants. J Dev Behav Pediatr 54. Harper DG, Stopa EG, McKee AC, et al. Differential circadian 2002;23:410-5. rhythm disturbances in men with Alzheimer disease and frontotem- 73. Harrison Y. The relationship between daytime exposure to light and poral degeneration. Arch Gen Psychiatry 2001;58:353-60. night-time sleep in 6-12-week-old infants. J Sleep Res 2004;13:345- 55. Hatfield CF, Herbert J, van Someren EJ, Hodges JR, Hastings MH. 52. Disrupted daily activity/rest cycles in relation to daily cortisol 74. Korte J, Hoehn T, Siegmund R. Actigraphic recordings of activity- rhythms of home-dwelling patients with early Alzheimer’s demen- rest rhythms of neonates born by different delivery modes. Chrono- tia. Brain 2004;127:1061-74. biol Int 2004;21:95-106. 56. Paavilainen P, Korhonen I, Lotjonen J, et al. Circadian activ- 75. Gossel-Symank R, Grimmer I, Korte J, Siegmund R. Actigraphic ity rhythm in demented and non-demented nursing-home residents monitoring of the activity-rest behavior of preterm and full-term measured by telemetric actigraphy. J Sleep Res 2005;14:61-8. infants at 20 months of age. Chronobiol Int 2004;21:661-71. 57. Greco KE, Deaton C, Kutner M, Schnelle JF, Ouslander JG. Psycho- 76. Korte J, Wulff K, Oppe C, Siegmund R. Ultradian and circadian active medications and actigraphically scored sleep quality in frail activity-rest rhythms of preterm neonates compared to full-term ne- nursing home patients. J Am Med Dir Assoc 2004;5:223-7. onates using actigraphic monitoring. Chronobiol Int 2001;18:697- 58. Ruths S, Straand J, Nygaard HA, Bjorvatn B, Pallesen S. Effect of 708. antipsychotic withdrawal on behavior and sleep/wake activity in 77. Yaron M, Lindgren K, Halbower AC, Weissberg M, Reite M, Nier- nursing home residents with dementia: a randomized, placebo-con- meyer S. Sleep disturbance after rapid ascent to moderate altitude trolled, double-blinded study. The Bergen District Nursing Home among infants and preverbal young children. High Alt Med Biol Study. J Am Geriatr Soc 2004;52:1737-43. 2004;5:314-20. 59. Fetveit A, Skjerve A, Bjorvatn B. Bright light treatment improves 78. Nishihara K, Horiuchi S, Eto H, Uchida S. The development of in- sleep in institutionalised elderly--an open trial. Int J Geriatr Psychia- fants’ circadian rest-activity rhythm and mothers’ rhythm. Physiol try 2003;18:520-6. Behav 2002;77:91-8. 60. Serfaty M, Kennell-Webb S, Warner J, Blizard R, Raven P. Double 79. Gay CL, Lee KA, Lee SY. Sleep patterns and fatigue in new moth- blind randomised placebo controlled trial of low dose melatonin for ers and fathers. Biol Res Nurs 2004;5:311-8. sleep disorders in dementia. Int J Geriatr Psychiatry 2002;17:1120- 80. El-Sheikh M, Buckhalt JA. Vagal regulation and emotional intensi- 7. ty predict children’s sleep problems. Dev Psychobiol 2005;46:307- 61. Tractenberg RE, Singer CM, Cummings JL, Thal LJ. The Sleep Dis- 17. orders Inventory: an instrument for studies of sleep disturbance in 81. Wolfson AR, Carskadon MA, Acebo C, et al. Evidence for the va- persons with Alzheimer’s disease. J Sleep Res 2003;12:331-7. lidity of a sleep habits survey for adolescents. Sleep 2003;26:213- 62. Alessi CA, Martin JL, Webber AP, Cynthia Kim E, Harker JO, Jo- 6.

SLEEP, Vol. 30, No. 4, 2007 528 Practice Parameters for the Use of Actigraphy—Morgenthaler et al 82. Amin R, Bean J, Burklow K, Jeffries J. The relationship between 11. McCurry SM, Logsdon RG, Vitiello MV, Teri L. Treatment of sleep sleep disturbance and pulmonary function in stable pediatric cystic and nighttime disturbances in Alzheimer’s disease: a behavior man- fibrosis patients. Chest 2005;128:1357-63. agement approach. Sleep Med 2004;5:373-7. 83. De Leersnyder H, De Blois MC, Claustrat B, et al. Inversion of the 12. Middleton B, Stone BM, Arendt J. Human circadian phase in 12:12 circadian rhythm of melatonin in the Smith-Magenis syndrome. J h, 200: <8 lux and 1000: <8 lux light-dark cycles, without scheduled Pediatr 2001;139:111-6. sleep or activity. Neurosci Lett 2002;329:41-4. 84. Wee R, Van Gelder RN. Sleep disturbances in young subjects with 13. Mongrain V, Lavoie S, Selmaoui B, Paquet J, Dumont M. Phase visual dysfunction. Ophthalmology 2004;111:297-302; discussion relationships between sleep-wake cycle and underlying circadian 302-3. rhythms in Morningness-Eveningness. J Biol Rhythms 2004;19:248- 85. Van der Heijden KB, Smits MG, Van Someren EJ, Gunning WB. 57. Idiopathic chronic sleep onset insomnia in attention-deficit/hyper- 14. Monk TH, Buysse DJ, Kennedy KS, Pods JM, DeGrazia JM, activity disorder: a circadian rhythm sleep disorder. Chronobiol Int Miewald JM. Measuring sleep habits without using a diary: the 2005;22:559-70. sleep timing questionnaire. Sleep 2003;26:208-12. 86. Paavonen EJ, Nieminen-von Wendt T, Vanhala R, Aronen ET, von 15. O’Reardon JP, Ringel BL, Dinges DF, et al. Circadian eating Wendt L. Effectiveness of melatonin in the treatment of sleep distur- and sleeping patterns in the night eating syndrome. Obes Res bances in children with Asperger disorder. J Child Adolesc Psycho- 2004;12:1789-96. pharmacol 2003;13:83-95. 16. Provini F, Albani F, Vetrugno R, et al. A pilot double-blind placebo- 87. Crabtree VM, Ivanenko A, Gozal D. Clinical and parental assess- controlled trial of low-dose pramipexole in sleep-related eating dis- ment of sleep in children with attention-deficit/hyperactivity disor- order. Eur J Neurol 2005;12:432-6. der referred to a pediatric sleep medicine center. Clin Pediatr (Phila) 17. Regestein QR, Friebely J, Shifren JL, et al. Self-reported sleep in 2003;42:807-13. postmenopausal women. Menopause 2004;11:198-207. 88. Montgomery-Downs HE, Crabtree VM, Gozal D. Actigraphic re- 18. Richards KC, Sullivan SC, Phillips RL, Beck CK, Overton-McCoy cordings in quantification of periodic leg movements during sleep AL. The effect of individualized activities on the sleep of nurs- in children. Sleep Med 2005;6:325-32. ing home residents who are cognitively impaired: a pilot study. J 89. Paavonen EJ, Fjallberg M, Steenari MR, Aronen ET. Actigraph Gerontol Nurs 2001;27:30-7. placement and sleep estimation in children. Sleep 2002;25:235-7. 19. Richards KC, Beck C, O’Sullivan PS, Shue VM. Effect of indi- 90. Fallone G, Seifer R, Acebo C, Carskadon MA. How well do school- vidualized social activity on sleep in nursing home residents with aged children comply with imposed sleep schedules at home? Sleep dementia. J Am Geriatr Soc 2005;53:1510-7. 2002;25:739-45. 20. Sadeh A, Keinan G, Daon K. Effects of stress on sleep: the moderat- ing role of coping style. Health Psychol 2004;23:542-5. 21. Shibui K, Uchiyama M, Iwama H, Ozaki S, Takahashi K, Okawa Articles not cited in the paper M. Periodic fatigue symptoms due to desynchronization in a pa- 1. Benson K, Friedman L, Noda A, Wicks D, Wakabayashi E, Yesav- tient with non-24-h sleep-wake syndrome. Psychiatry Clin Neurosci age J. The measurement of sleep by actigraphy: direct comparison 1998;52:477-81. of 2 commercially available actigraphs in a nonclinical population. 24. Tuisku K, Holi MM, Wahlbeck K, Ahlgren AJ, Lauerma H. Quan- Sleep 2004;27:986-9. titative rest activity in ambulatory monitoring as a physiological 2. Carney CE, Lajos LE, Waters WF. Wrist actigraph versus self-report marker of restless legs syndrome: a controlled study. Mov Disord in normal sleepers: sleep schedule adherence and self-report valid- 2003;18:442-8. ity. Behav Sleep Med 2004;2:134-43; discussion 144-7. 25. Tuisku K, Holi MM, Wahlbeck K, Ahlgren AJ, Lauerma H. Actom- 3. Denise P, Bocca ML. Effects of zolpidem 10 mg, zopiclone 7.5 mg etry in measuring the symptom severity of restless legs syndrome. and flunitrazepam 1 mg on night-time motor activity. Eur Neuropsy- Eur J Neurol 2005;12:385-7. chopharmacol 2003;13:111-5. 26. Tworoger SS, Davis S, Vitiello MV, Lentz MJ, McTiernan A. Fac- 4. Fontana Gasio P, Krauchi K, Cajochen C, et al. Dawn-dusk simulators associated with objective (actigraphic) and subjective sleep tion light therapy of disturbed circadian rest-activity cycles in de- quality in young adult women. J Psychosom Res 2005;59:11-9. mented elderly. Exp Gerontol 2003;38:207-16. 27. Zucconi M, Oldani A, Castronovo C, Ferini-Strambi L. Cabergo- 5. Guilleminault C, Palombini L, Poyares D, Chowdhuri S. Chronic line is an effective single-drug treatment for restless legs syndrome: insomnia, premenopausal women and sleep disordered breathing: clinical and actigraphic evaluation. Sleep 2003;26:815-8. part 2. Comparison of nondrug treatment trials in normal breathing 28. Nelson J, Harvey AG. Pre-sleep imagery under the microscope: a and UARS post menopausal women complaining of chronic insom- comparison of patients with insomnia and good sleepers. Behav Res nia. J Psychosom Res 2002;53:617-23. Ther 2003;41:273-84. 6. Guilleminault C, Palombini L, Poyares D, Chowdhuri S. Chronic insomnia, postmenopausal women, and sleep disordered breathing: part 1. Frequency of sleep disordered breathing in a cohort. J Psy- chosom Res 2002;53:611-5. 7. Guilleminault C, Moscovitch A, Yuen K, Poyares D. Atypical sexual behavior during sleep. Psychosom Med 2002;64:328-36. 8. Hoekert M, der Lek RF, Swaab DF, Kaufer D, Van Someren EJ. Comparison between informant-observed and actigraphic assessments of sleep-wake rhythm disturbances in demented residents of homes for the elderly. Am J Geriatr Psychiatry 2006;14:104-11. 9. Martin JL, Mory AK, Alessi CA. Nighttime oxygen desaturation and symptoms of sleep-disordered breathing in long-stay nursing home residents. J Gerontol A Biol Sci Med Sci 2005;60:104-8. 10. Martin JL, Webber AP, Alam T, Harker JO, Josephson KR, Alessi CA. Daytime sleeping, sleep disturbance, and circadian rhythms in the nursing home. Am J Geriatr Psychiatry 2006;14:121-9.

SLEEP, Vol. 30, No. 4, 2007 529 Practice Parameters for the Use of Actigraphy—Morgenthaler et al