Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 1.
Published in final edited form as: J Sleep Res. 2019 Apr 2;28(4):e12854. doi: 10.1111/jsr.12854

Relation between ambulatory actigraphy and laboratory polysomnography in insomnia practice and research

Dana Withrow 1, Thomas Roth 1,2, Gail Koshorek 1, Timothy Roehrs 1,2
PMCID: PMC6639797  NIHMSID: NIHMS1041225  PMID: 30941838

Abstract

Actigraphy is increasingly used in practice and research studies because of its relative low cost and decreased subject burden. How multiple nights of at-home actigraphy compare to one independent night of in-laboratory polysomnography (PSG) has not been examined in people with insomnia. Using event markers (MARK) to set time in bed (TIB) compared to automatic program analysis (AUTO) has not been systematically evaluated. Subjects (n = 30) meeting DSM-5 criteria for insomnia and in-laboratory PSG sleep efficiency (SE) of <85% were studied. Subjects were free of psychiatric, sleep or circadian disorders, other chronic conditions and medications that effect sleep. Subjects had an in-laboratory PSG, then were sent home for 7 nights with Philips Actiwatch Spectrum Plus. Data were analysed using Philips Actiware version 6. Using the mean of seven nights, TIB, total sleep time (TST), SE, sleep-onset latency (SOL) and wake after sleep onset (WASO) were examined. Compared to PSG, AUTO showed longer TIB and TST and less WASO. MARK only differed from PSG with decreased WASO. Differences between the PSG night and the following night at home were found, with better sleep on the first night home. Actigraphy in people with insomnia over seven nights is a valid indicator of sleep compared to an independent in-laboratory PSG. Event markers increased the validity of actigraphy, showing no difference in TIB, TST, SE and SOL. AUTO was representative of SE and SOL. Increased SE and TST without increased TIB suggests possible compensatory sleep the first at night home after in-laboratory PSG.

Keywords: actigraphy, event markers, insomnia, polysomnography, sleep

1 ∣. INTRODUCTION

Actigraphy provides clinicians and researchers with a way to examine sleep patterns at home in a non-invasive, economical manner, over an extended period of time when compared to the reference standard of laboratory polysomnography (PSG). Barriers that may reduce the possibility of an in-laboratory PSG for certain patients (e.g., financial, availability of sleep laboratory, health, etc.) necessitate other methods for assessing sleep clinically. In research settings where multiple nights over extended time periods are required or large sample sizes are desired, in-laboratory PSG is less feasible. Although in-laboratory PSG provides complete information regarding sleep and wake time, sleep stages and primary sleep disorders, the value of actigraphy is its potential for natural at-home assessment of sleep and its timing. Actigraphy is not recommended to be used as a replacement for PSG, but it is increasingly used as an alternative measure to in-laboratory PSG when PSG is not clinically viable or when studying large research populations (Ancoli-Israel et al., 2003; Morgenthaler et al., 2007; Sadeh, 2011). The question is how reliable and valid actigraphy is relative to in-laboratory PSG and for which patient or research study populations.

Actigraphy taken simultaneously with PSG has been shown to be valid and reliable in healthy sleepers, measuring TST (total sleep time) and SE (sleep efficiency) with high sensitivity (>90%). These high correlations are because of the ability of actigraphy to correctly detect sleep periods (high sensitivity) (Ancoli-Israel et al., 2003; Jean-Louis et al., 1996; Kosmadopoulos, Sargent, Darwent, Zhou, & Roach, 2014; Weiss, Johnson, Berger, & Redline, 2010). When sleep becomes more disturbed (e.g., insomnia), these findings are not as robust. However, studies have validated the use of actigraphy in individuals with disturbed sleep reporting moderate correlations to simultaneous in-laboratory PSG (Hauri & Wisbey, 1992; Kushida et al., 2001; Lichstein et al., 2006; Natale, Plazzi, & Martoni, 2009; Paquet, Kawinska, & Carrier, 20072007). The reduced validity in people with insomnia is likely to be because of the difficulty actigraphy has in detecting periods of wake (low specificity), both before and during sleep, which are elevated in insomnia subjects relative to healthy normal subjects (Marino et al., 2013; Paquet et al., 2007; Sadeh et al., 2011). In part, the problem of wake detection in insomnia is in establishing the timing of the sleep period to be analysed, based on the subject’s intent to sleep and arise.

In sleep laboratories the initiation and conclusion of the sleep period is defined by the technician marking lights off/on. In the at-home situation, as with actigraphy, the beginning and the end of the sleep period are dependent on the subject identifying them. In the case of some actigraphy systems, there is an event marker that the subject is supposed to activate at the beginning and end of the sleep period, to indicate time-in-bed (TIB). Given that event markers are not routinely utilized, software has been developed to determine the beginning and the end of the sleep period with actigraphy use. The impact these “event markers” have on the accuracy of actigraphy data has not been determined systematically in insomnia. This study will evaluate the potential differences that use of event markers may have in sleep measures derived from actigraphy.

Night-to-night variability in the severity of sleep disturbances among individuals with insomnia is not comprehensively captured by a single in-laboratory PSG (Newell, Mairesse, Verbanck, & Neu, 2012). Although sleep logs may fill in gaps of knowledge, it is known that the subjective measures of sleep vary considerably from objective PSG and typically overestimate the severity of the objectively determined sleep disturbance (Kushida et al., 2001; McCall & Edinger, 1992; McCall & McCall, 2012). Actigraphy provides an opportunity to objectively evaluate variability in sleep parameters over time, which is especially important in people with insomnia as they are known to have variable sleep from night to night (Buysse et al., 2010; Perlis et al., 2014; Rowe et al., 2008; Van Someren, 2007).

Previous studies examining comparisons of actigraphy to PSG have measured both PSG and actigraphy simultaneously to determine validity and have been carried out in multiple medical conditions. In such studies the method of assessment (PSG versus wrist-worn device) is compared, but not the sleep environment (inlaboratory versus at home). No studies have compared sleep measures from multiple nights of at-home actigraphy to one independent night of in-laboratory PSG in insomnia. The decreased correlations found between concurrent PSG and actigraphy in individuals with insomnia brings into question the validity of using actigraphy as a substitute measure for PSG (Pollak, Tryon, Nagaraja, & Dzwonczyk, 2001). However, in order to interpret research and findings using actigraphy and compare outcomes across different data collection methodologies, it is necessary to understand the specific parameter differences between laboratory PSG and at-home actigraphy.

Currently, recommendations for use of ambulatory actigraphy in individuals with insomnia are limited to characterizing sleep duration and circadian patterns (Morgenthaler et al., 2007). Another important question regarding the place of actigraphy in insomnia research and practice is its sensitivity to changes in sleep, because of changes in sleep environment, disruptive effects of sleep monitoring techniques or therapeutic interventions. Actigraphy has been shown to detect differences in measuring response to therapeutic interventions in insomnia (Vallières & Morin, 2003). However, past studies are equivocal regarding the ability of actigraphy to detect “first night effects” on the first night following PSG. Blackwell et al. have reported on at-home PSG followed by three nights of actigraphy assessed in older men and found there was recovery sleep on the night following at-home PSG (Blackwell, Paudel, Redline, Ancoli-Israel, & Stone, 2017). On the other hand, McCall & McCall analysed actigraphy data from a night of a PSG and the week prior and week following in-laboratory PSG and found the possibility of a “reverse” first night effect (i.e., better in-laboratory sleep) in people with insomnia (McCall & McCall, 2012). One approach to further evaluate actigraphic sensitivity to changes in sleep environment in people with insomnia is to evaluate sleep at home after the first night in the sleep laboratory. Thus, investigation into a possible “rebound” sleep the night following the first night of in-laboratory PSG is warranted, as it also tests the sensitivity of actigraphy.

This study aims to examine the strengths and weaknesses of multiple nights of actigraphy in people with insomnia and to determine the relation between sleep measures collected during a single independent night of in-laboratory PSG, considered the reference standard for documenting sleep disturbance, versus multiple nights of at-home actigraphy. Further, the study will assess the added value of the use of event markers to establish the sleep period and will assess the sensitivity of actigraphy in detection of rebound sleep following in-laboratory PSG.

2 ∣. MATERIALS AND METHODS

2.1 ∣. Participants and settings

Data were collected from individuals who were recruited for a 6-month, on-going pharmacological intervention insomnia research study. Participants came from multiple sources, including Henry Ford Hospital patients via flyer or referral, Henry Ford Hospital employees via a daily news email advertisement and others from online informational sources or friend referral. Interested participants underwent a telephone screening, physical appointment and one overnight in-laboratory PSG screen.

This study was approved by the Henry Ford Hospital Institutional Review Board and maintained approval from a data safety monitoring board throughout. Informed consent was given at the first clinic appointment, which was followed by a blood draw, urine test for drugs or pregnancy, physical examination and medical history taken by a physician, and multiple questionnaires. All individuals met the DSM 5 insomnia criteria (American Psychiatric Association, 2013).

After the single nocturnal screening PSG and a multiple sleep latency test (MSLT) the following day, subjects were instructed on how to use an Actiwatch (Spectrum Plus, Philips Healthcare, Bend, OR) and sent home with the instruction to wear the device for seven consecutive days and nights beginning on the first night home.

Participants (n = 30) age ranged from 24 to 62 years and they were in general in good health (Table 1). Individuals with a body mass index (BMI) > 38, having acute or unstable illnesses that have the potential to disturb sleep, chronic illnesses, current or past year psychiatric disease, or history of alcohol or substance abuse in the last 2 years, consuming >14 standard drinks/week, smoking during the night (23:00–07:00 hours) and taking medications including anxiolytics, hypnotics, antidepressants, sedating H1 antihistamines, respiratory stimulants, diet aids or narcotic analgesics were excluded.

TABLE 1.

Demographics

All subjects (n = 30)
Age 45.1
Sex (female) 25 (83.33%)
Race (Caucasian) 10 (33.33%)
Race (African American) 17 (56.67%)
Race (Asian) 3 (10.00%)
ESS 7.13
ISI 16.8
PSG TST (min) 364.4
PSG SE % 75.9
PSG SOL (min) 29.2
PSG WASO (min) 92.2
Open in a new tab

ESS: Epworth Sleepiness Scale; ISI: Insomnia Severity Index; PSG: polysomnography; TST: total sleep time; SE: sleep efficiency; SOL: sleep-onset latency; WASO: wake after sleep onset.

In addition, subjects underwent a screening PSG to rule out any sleep disorders other than insomnia, including: sleep disordered breathing, periodic leg movements, narcolepsy or circadian disorders. Importantly, participants included in this study were required to have a sleep efficiency (SE) (total sleep time/time in bed) of 85% or less. This cut-off was used to demonstrate that participants have objectively measured disturbed sleep in addition to meeting the DSM 5 criteria for insomnia (Randall, Roehrs, & Roth, 2012).

2.2 ∣. Actigraphy

Subjects were instructed on the use of the Actiwatch Spectrum Plus (Philips Healthcare, Bend, OR) and to begin wearing the device the first night at home. This particular actigraphy device uses a micro electro-mechanical system (MEMS)-type accelerometer with a sampling rate of 32 Hz. Parameters were set to a 30-s epoch length and medium sensitivity and both activity and marker selections were set at the default settings. Instructions were to wear the actigraph day and night on their non-dominant wrist at all times except when submerged in water (e.g., bathing or swimming). At night, when they were in bed and were ready to try to fall asleep, they were told to use the event marker on the actigraph to mark their bedtime. In the morning, when subjects woke up to start their day, they were instructed to mark the actigraph again. Specifically, the instruction read to them was: “at night, when you get in bed to try and sleep, make sure to press the left button for 3 s (a flashing border will appear on the screen to confirm the action) to mark your bedtime. Please do the same when you get out of bed to start your day.” Importantly, if subjects forgot to mark the actigraph and more than 10 min had passed, they were told to not mark the actigraph and instead write down the time they meant to press it so the mark could be entered into the software manually. Participants were instructed to spend between 7 and 9 hr in bed during the week.

Philips Actiware version 6 software was used in the analysis of the actigraph data. Sleep parameters were derived using both the Actiware version 6 software automatic determination of time in bed (AUTO) and subject-marked time in bed based on the event markers (MARK).

2.3 ∣. Polysomnography

For the single-night screening PSG, subjects were instructed to report to the laboratory approximately 2 hr before their normal bedtime. Normal bedtime was determined by a 1-week sleep diary completed by the subject during the week leading up to their PSG appointment. An 8-hr PSG was conducted, which included standard central (C3-A2) and occipital (Oz-A2) electroencephalograms (EEGs), bilateral horizontal electro-oculograms (EOG), submental electromyogram (EMG) and electrocardiogram (ECG). An electrode placed over the left tibialis muscle monitored leg movements. Respiration was measured by a nasal thermistor. Subjects with respiratory disturbances (apnea-hypopnea index [AHI] > 10) or with periodic limb movement arousal indices (PLMAI) > 10/hr were excluded from the study. Subjects were required to demonstrate a screening sleep efficiency of less than or equal to 85% and have no other sleep disorders. Scoring of records was completed using standard Rechtschaffen and Kales methods via 30-s epochs (Rechtschaffen & Kales, 1968) by scorers who maintained a ≥90% scoring reliability.

2.4 ∣. Analyses

Time in bed (TIB), total sleep time (TST), sleep efficiency (SE), sleep-onset latency (SOL) and wake after sleep onset (WASO) were derived from the Philips Actiware version 6 software analyses of the actigraphic data for each of the seven nights. These five actigraphic parameters were derived using the Phillips automated software (AUTO) and the subject-marked (MARK) software analyses. Then means for the actigraphic data for both AUTO and MARK data were determined.

Differences in minutes of PSG TST and WASO versus the mean actigraphic-derived TST and WASO using AUTO and MARK data were determined using Bland-Altman plots. Additionally, the actigraph means for the five parameters were compared to the same PSG parameters using paired t tests. First, one set of tests compared the AUTO-derived data to PSG and one set compared the MARK-derived data to PSG (see Table 2). Next, the five MARK versus AUTO-derived actigraph parameters were also compared to each other (see Table 3). The first-night at-home actigraph data (AUTO and MARK) were compared to the PSG data (see Table 4) and finally the first-night actigraph data (AUTO and MARK) were compared to the mean of nights 2–7 (see Tables 5 and 6). To correct for family-wise multiple comparisons, alpha was corrected using the Holm-Bonferroni method.

TABLE 2.

Laboratory polysomnography night compared to automatic program analysis (AUTO) and event markers (MARK) seven nights actigraphy data (n = 21)

PSG Seven-night
AUTO average		 Seven-night
MARK average
TIB (min) 480.0 (0.0) 515.2 (13.45)a 486.7 (11.62)
TST (min) 376.3 (7.69) 425.3 (12.21)b 394.9 (11.63)
SE (%) 78.3 (1.60) 81.9 (1.08) 81.6 (1.54)
SOL (min) 30.9 (7.64) 29.0 (3.76) 30.4 (7.04)
WASO (min) 81.7 (6.73) 44.2 (3.58)b 38.7 (3.35)b
Open in a new tab
a

Mean (SE) versus PSG p < 0.05,

b

Mean (SE) versus PSG p < 0.01. PSG: polysomnography; TIB, time in bed; TST: total sleep time; SE: sleep efficiency; SOL: sleep-onset latency; WASO: wake after sleep onset.

TABLE 3.

Averages of seven nights of automatic program analysis (AUTO) and event markers (MARK) actigraphy (n = 21)

Seven-night AUTO
average		 Seven-night MARK
average		 AUTO/MARK difference
TIB (min) 515.2 (13.45) 486.7(11.62) 28.5a
TST (min) 425.3 (12.21) 394.9 (11.63) 30.4b
SE (%) 81.9 (1.08) 81.6 (1.54) 0.3
SOL (min) 29.0 (3.76) 30.4 (7.04) −1.4
WASO (min) 44.2 (3.58) 38.7 (3.35) 5.5
Open in a new tab
a

Mean (SE) versus p < 0.05

b

Mean (SE) versus p < 0.001. TIB, time in bed; TST: total sleep time; SE: sleep efficiency; SOL: sleep-onset latency; WASO: wake after sleep onset.

TABLE 4.

Laboratory polysomnography (PSG) night compared to actigraphy on first night home following PSG (n = 21)

PSG Night-1 AUTO Night-1 MARK
TIB (min) 480.0 (0.0) 544.0 (31.16) 509.0 (22.44)
TST (min) 376.3 (7.69) 474.6 (27.29)a 432.8 (17.67)a
SE (%) 78.3 (1.60) 86.2 (1.64)b 86.1 (1.64)b
SOL (min) 30.9 (7.64) 17.8 (5.79) 17.1 (7.27)
WASO (min) 81.7 (6.73) 46.4 (6.64)b 44.8 (5.61)b
Open in a new tab
a

Mean (SE) versus PSG p < 0.01

b

Mean (SE) versus PSG p < 0.001. AUTO: automatic program analysis; MARK: event markers; TIB, time in bed; TST: total sleep time; SE: sleep efficiency; SOL: sleep-onset latency; WASO: wake after sleep onset.

TABLE 5.

First night of automatic program analysis (AUTO) actigraphy compared to AUTO on Nights 2–7 (n = 21)

Night-1 AUTO Night-2–7 AUTO Night 1/Night
2–7 difference
TIB (min) 544.0 (31.16) 510.3 (12.68) 33.7
TST (min) 474.6 (27.29) 416.7 (11.43) 57.9a
SE (%) 86.2 (1.64) 81.2 (1.08) 5.0b
SOL (min) 17.8 (5.79) 30.9 (4.15) −13.1
WASO (min) 46.4 (6.64) 43.8 (3.36) 2.6
Open in a new tab
a

Mean (SE) versus Night 1 p < 0.02,

b

Mean (SE) versus Night 1 p < 0.002. TIB, time in bed; TST: total sleep time; SE: sleep efficiency; SOL: sleep-onset latency; WASO: wake after sleep onset.

TABLE 6.

First night of event markers (MARK) actigraphy compared to MARK on Nights 2–7 (n = 21)

Night-1 MARK Night-2–7 MARK Night 1/Nights
2–7 difference
TIB (min) 509.0 (22.44) 479.5 (14.13) 29.5
TST (min)a 432.8 (17.67) 385.8 (12.75) 47.0
SE (%) 86.1 (1.64) 80.9 (1.62) 5.2b
SOL (min) 17.1 (7.27) 32.7 (7.97) −15.6
WASO (min) 44.8 (5.61) 36.6 (3.63) 8.2
Open in a new tab
a

Mean (SE) versus Night 1 p < 0.02,

b

Mean (SE) versus Night 1 p < 0.001. TIB, time in bed; TST: total sleep time; SE: sleep efficiency; SOL: sleep-onset latency; WASO: wake after sleep onset.

3 ∣. RESULTS

Twenty-one of the 30 participants had sufficient data to analyse both automatic and event-marked data (i.e., >2 marked nights). Averages available over the seven-night period were calculated with the requirement that at least three nights of marked data were available for a given participant. Out of a possible total of 210 (30 participants × 7 nights) nights of data, 45 nights were missing one or both event markers, indicating that 21% of nights were excluded as a result of subject non-compliance. Thus, the study “n” in all analyses is n = 21 (i.e., had at least three nights of marked data to contribute to the means).

3.1 ∣. Automatic actigraphy analysis

When compared to PSG, AUTO actigraphy reported significantly longer TIB (t = −2.56, p = 0.02) and TST (t =−3.19, p < 0.01) and shorter WASO (t = 4.75, p < 0.001). There was no difference found for SE and SOL between in-laboratory PSG and AUTO actigraphy over seven nights (Figure 1; Table 2). Bland-Altman analysis, which is typically used to evaluate agreement between two measures, showed longer AUTO actigraphy TST compared to PSG by 49 min (p < 0.01).

FIGURE 1.

FIGURE 1

Open in a new tab

Bland-Altman plot of polysomnography (PSG) versus actigraphy automatic program analysis (AUTO) total sleep time. Comparison of PSG total sleep time (TST) and AUTO TST over averages of seven nights. AUTO actigraphy reported longer TST than PSG by 49 min (p < 0.05). n = 21

3.2 ∣. Event marker actigraphy analysis

Event marker data (MARK) when compared to in-laboratory PSG showed a significant difference only for WASO; MARK reported less WASO than PSG (t = 6.27, p < 0.001). There was no significant difference for TIB, TST, SE or SOL between MARK and PSG (Figure 2; Table 2). Bland-Altman analysis shows that MARK actigraphy TST compared to PSG was greater by 18.6 min, but this was not statistically significant (p = 0.24).

FIGURE 2.

FIGURE 2

Open in a new tab

Bland-Altman plot of polysomnography (PSG) versus actigraphy event markers (MARK) total sleep time. Comparison of PSG total sleep time (TST) and MARK TST over averages of seven nights. MARK actigraphy reported longer TST than PSG by 18.6 min (not significant, p = 0.24). n = 21

Comparing results on nights with event marker versus nights without marker, AUTO reported significantly longer TIB (t = 2.55, p = 0.02) and TST (t = 3.93, p < 0.001) than MARK.

3.3 ∣. First-night effects

AUTO, TST (t = 3.45, p < 01) and SE (t = 3.39, p < 0.01) on the first night home were elevated compared to the PSG laboratory night, whereas WASO (t = 3.75, p < 0.001) was significantly lower (Table 4). Given that the seven-night AUTO analysis also showed increased TIB and TST and decreased WASO on the first night home, SE was the only additional parameter showing an actigraph versus PSG difference.

MARK also showed significantly higher TST (t = 2.70, p < 0.01) and SE (3.35, p < 0.01) on the first night home when compared to PSG. Night-1 MARK WASO (t = 3.94, p < 0.001) was significantly less than PSG-measured WASO (Table 4). Findings from the seven-night MARK analysis previously showed less WASO, making TST and SE both emerging differences on the first night home.

Finally, both AUTO (t = 3.39, p < 0.01) and MARK (t = 2.59, p < 0.02) on Night 1 showed significantly longer TST and increased SE (AUTO: t = 3.45, p < 0.01; MARK: t = 3.62, p < 0.001) compared to nights 2–7.

4 ∣. DISCUSSION

This study examined multiple-nights at-home actigraphy versus a single independent laboratory PSG night for assessing sleep in otherwise healthy subjects with insomnia. Examining this relation with and without event markers provides guidelines for interpreting actigraphy findings compared to what a single independent night PSG might be expected to show. Seven nights of actigraphy using AUTO analysis was found to show longer TST, TIB and less WASO when compared to a single independent night of laboratory PSG. These findings are similar to past studies reporting that concurrent actigraphy tends to overestimate TST and underestimate WASO in individuals with insomnia (although the extent varies based on scoring methods) (Friedman et al., 2000; Kanady, Drummond, & Mednick, 2011). However, when examining seven nights of MARK data versus PSG, WASO was the only significant difference, with actigraphy reporting less WASO by 43 min. Event markers (MARK) minimized the difference in TST and TIB detected by the AUTO algorithm, reducing TST differences from 49 min to 18.6 min.

Overall these data suggest that sleep measures (excluding WASO) when using event-marked actigraphy at home for seven nights show similar findings to a single independent night of laboratory PSG. Automatic analysis of actigraphy data showed a tendency to overestimate TST and TIB when compared to laboratory PSG (however, event markers minimize this effect) but still showed similar measures in SOL and SE.

In regards to evaluation of a laboratory first-night effect measured by actigraphy on the first night home immediately after the PSG night, MARK data showed an increase in TST and SE. This increase in SE from PSG (78.3%) to AUTO (86.2%) and MARK (86.1%) indicates that there may be compensatory sleep on the first night home after a first night in the laboratory. Importantly, the TIB indicated by MARK data did not differ significantly compared to the PSG night, showing that subjects were not in bed any longer than the PSG night but still experienced increased TST and SE. This finding is supported by AUTO and MARK Night-1 data showing TST and SE that were significantly higher than Nights 2–7.

Actigraphy using the AUTO data showed longer TIB than PSG. Importantly, MARK data did not, suggesting that subjects instructed to spend 7–9 hr in bed followed these guidelines and were in bed for a similar time to the mandated 8-hr PSG. This finding supports the utility of actigraphy as a reliable measure of study instruction compliance, such as having adequate time in bed, especially with the use of event markers.

Previous studies evaluating the validity of actigraphy relative to in-laboratory PSG employed simultaneous actigraph and PSG recordings. There are two difficulties with this approach. First, this is not the way at-home actigraphy is executed in the literature, where actigraph-based studies do not have a PSG component. Secondly, the use of simultaneous recording confers a bias towards showing concordance between the two methods. In-laboratory PSG typically sets the initiation and conclusion of the PSG recording period, which the AUTO versus MARK analysis of this study showed to be a weakness in achieving PSG and actigraphy concordance.

Our results suggest that MARK actigraphy has an intermediate place in insomnia practice and research. Diary-reported sleep has been used as an alternative to in-laboratory PSG, but dairies are subject to bias and dependent on recall. Actigraphy provides an objective realtime assessment of sleep in the patient’s at-home sleep environment. Marking provides a relatively good estimate of sleep duration and timing. Studies show the morbidity associated with insomnia is found primarily among those patients with DSM-V diagnosed insomnia and PSG short sleep. Finally, here we have shown MARK actigraphy is sensitive to a change in sleep environment (i.e., from the laboratory to home), which shows its potential in both clinic and research settings.

There are limitations that must be recognized as to the generalizability to clinical insomnia populations seen in general medical practice. Subjects were relatively healthy with no psychiatric disorders and not taking any medications that have CNS effects. Past studies have shown that mental and physical health disorders such as depression or chronic pain disturb sleep versus healthy normal subjects. The known comorbidity of insomnia and other mental or physical health disorders brings into question the generalizability of the current results to a broader insomnia population.

This study included only insomnia patients with PSG-documented disturbed sleep (i.e., SE < 85%). It is possible that those not meeting this criterion (i.e., having better SE) might produce differing results. However, rather than considering this a study limitation, it may be a strength. In fact, using an insomnia population with objective sleep disturbance better tests the limits of actigraphy versus PSG. Given the low specificity of actigraphy in detecting wake, a population with greater amounts of wake provides a best-case test of the limits of actigraphy.

Of the 30 subjects from whom data were collected, only 21 remembered to use event markers on the first night home, although over all the nights only 21% were lost as a result of failure to mark. This limited sample size may be another limitation of the study; future studies with an increased number of participants should be conducted to determine the reliability of these findings. This also raises the issue of subject compliance using event markers. Ideally, use of actigraphy and event marking should be carried out with emphasis on the importance of instructing patients to remember to use event markers. Of note, all nine subjects with compliance difficulties failed to mark on the first night home after the PSG, which suggests a need to specifically monitor first-night compliance.

Because this study only used Philips Actiwatch Spectrum Plus, this information cannot necessarily be generalized to other brands or actigraphy devices. Standardizing a specific sensitivity setting on actigraphy devices, as well as a recommended duration of home actigraphy use, would help in cross-study data comparisons.

In conclusion, based on these data, it appears MARK actigraphy is an intermediate alternative to PSG and diary-reported sleep both for practice and research. In this study, averages over seven nights of AUTO actigraphy showed longer TIB and TST and less WASO than laboratory PSG. However, event markers mitigated these differences; the only sleep measure that was significantly different to PSG was WASO, which was overestimated by 18 min on average. Subject instruction and use of event markers mitigate a potential systematic difference in actigraphy software analysis and are recommended in order to collect more accurate data. Further, actigraphy was sensitive to an improvement in sleep circumstances (i.e., sleep environment in the laboratory versus at home and intrusiveness of recording methods).

ACKNOWLEDGEMENTS

Supported by the NIDA, grant #: R01DA038177, awarded to Dr Timothy Roehrs. Thanks to the Henry Ford Sleep Disorders and Research Center technical staff.

Funding information

National Institute of Drug Abuse, Grant/Award Number: R01 DA038177

Footnotes

CONFLICT OF INTEREST

None of the authors have disclosures to make relevant to the current paper.

REFERENCES

  1. Ancoli-Israel S, Cole R, Alessi C, Chambers M, Moorcroft W, & Poliak CP (2003). The role of actigraphy in the study of sleep and circadian rhythms. Sleep, 26, 342–392. 10.1093/sleep/26.3.342 [DOI] [PubMed] [Google Scholar]
  2. American Psychiatric Association. 2013. Diagnostic and statistical manual of mental disorders (DSM-5®). Arlington, VA: American Psychiatric Pub. [Google Scholar]
  3. Blackwell T, Paudel M, Redline S, Ancoli-Israel S, & Stone KL (2017). A novel approach using actigraphy to quantify the level of disruption of sleep by in-home polysomnography: The MrOS Sleep Study Sleep disruption by polysomnography. Sleep Medicine, 32, 97–104. 10.1016/j.sleep.2016.11.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Buysse DJ, Cheng Y, Germain A, Moul DE, Franzen PL, Fletcher M, & Monk TH (2010). Night-to-night sleep variability in older adults with and without chronic insomnia. Sleep Medicine, 11, 56–64. 10.1016/j.sleep.2009.02.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Friedman L, Benson K, Noda A, Zarcone V, Wicks DA, O’Connell K, … Yesavage JA (2000). An actigraphic comparison of sleep restriction and sleep hygiene treatments for insomnia in older adults. Journal of Geriatric Psychiatry and Neurology, 13(1), 17–27. 10.1177/089198870001300103 [DOI] [PubMed] [Google Scholar]
  6. Hauri PJ, & Wisbey J (1992). Wrist actigraphy in insomnia. Sleep, 15, 293–297. 10.1093/sleep/15.4.293 [DOI] [PubMed] [Google Scholar]
  7. Jean-Louis G, Von Gizycki H, Zizi F, Fookson J, Spielman A, Nunes J, … Taub H (1996). Determination of sleep and wakefulness with the actigraph data analysis software (ADAS). Sleep, 19, 739–743. [PubMed] [Google Scholar]
  8. Kanady JC, Drummond SPA, & Mednick SC (2011). Actigraphic assessment of a polysomnographic-recorded nap: A validation study. Journal of Sleep Research, 20, 214–222. 10.1111/j.1365-2869.2010.00858.x [DOI] [PubMed] [Google Scholar]
  9. Kosmadopoulos A, Sargent C, Darwent D, Zhou X, & Roach GD (2014). Alternatives to polysomnography (PSG): A validation of wrist actigraphy and a partial-PSG system. Behavior Research Methods, 46, 1032–1041. 10.3758/s13428-013-0438-7 [DOI] [PubMed] [Google Scholar]
  10. Kushida CA, Chang A, Gadkary C, Guilleminault C, Carrillo O, & Dement WC (2001). Comparison of actigraphic, polysomnographic, and subjective assessment of sleep parameters in sleep-disordered patients. Sleep Medicine, 2, 389–395. 10.1016/s1389-9457(00)00098-8 [DOI] [PubMed] [Google Scholar]
  11. Lichstein KL, Stone KC, Donaldson J, Nau SD, Soeffing JP, Murray D, … Aguillard RN (2006). Actigraphy validation with insomnia. Sleep, 29, 232–239. [PubMed] [Google Scholar]
  12. Marino M, Li Y, Rueschman MN, Ellenbogen JM, Solet JM, Dulin H, … Buxton OM (2013). Measuring sleep: Accuracy, sensitivity, and specificity of wrist actigraphy compared to polysomnography. Sleep, 36, 1747–1755. 10.5665/sleep.3142 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. McCall WV, & Edinger JD (1992). Subjective total insomnia: An example of sleep state misperception. Sleep, 15, 71–75. [PubMed] [Google Scholar]
  14. McCall C, & McCall WV (2012). Objective vs subjective measurements of sleep in depressed insomniacs: First night effect or reverse first night effect? Journal of Clinical Sleep Medicine, 8, 59–65 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Morgenthaler T, Alessi C, Friedman L, Owens J, Kapur V, Boehlecke B, … Pancer J. (2007). Practice parameters for the use of actigraphy in the assessment of sleep and sleep disorders: An update for 2007. Sleep, 30, 519–529. 10.1093/sleep/30.4.519 [DOI] [PubMed] [Google Scholar]
  16. Natale V, Plazzi G, & Martoni M (2009). Actigraphy in the assessment of insomnia: A quantitative approach. Sleep, 32, 767–771. 10.1093/sleep/32.6.767 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Newell J, Mairesse O, Verbanck P, & Neu D (2012). Is a one-night stay in the lab really enough to conclude? First-night effect and night-to-night variability in polysomnographic recordings among different clinical population samples. Psychiatry Research, 200, 795–801. 10.1016/j.psychres.2012.07.045 [DOI] [PubMed] [Google Scholar]
  18. Paquet J, Kawinska A, & Carrier J (2007). Wake detection capacity of actigraphy during sleep. Sleep, 30, 1362–1369. 10.1093/sleep/30.10.1362 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Perlis ML, Zee J, Swinkels C, Kloss J, Morgan K, David B, & Morales K (2014). The incidence and temporal patterning of insomnia: A second study. Journal of Sleep Research, 23, 499–507. 10.1111/jsr.12150 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Pollak CP, Tryon WW, Nagaraja H, & Dzwonczyk R(2001). How accurately does wrist actigraphy identify the states of sleep and wakefulness? Sleep, 24, 957–965. 10.1093/sleep/24.8.957 [DOI] [PubMed] [Google Scholar]
  21. Randall S, Roehrs TA, & Roth T (2012). Efficacy of eight months of nightly zolpidem: A prospective placebo-controlled study. Sleep, 35, 1551–1557. 10.5665/sleep.2208 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Rechtschaffen A, Kales A; University of California Los Angeles Brain Information Service; NINDB Neurological Information Network (US). A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Washington, DC: United States Government Printing Office; 1968. [Google Scholar]
  23. Rowe M, McCrae C, Campbell J, Horne C, Tiegs T, Lehman B, & Cheng J (2008). Actigraphy in older adults: Comparison of means and variability of three different aggregates of measurement. Behavioral Sleep Medicine, 6, 127–145. 10.1080/15402000801952872 [DOI] [PubMed] [Google Scholar]
  24. Sadeh A (2011). The role and validity of actigraphy in sleep medicine: An update. Sleep Medicine Reviews, 15, 259–267. 10.1016/j.smrv.2010.10.001 [DOI] [PubMed] [Google Scholar]
  25. Vallières A, & Morin CM (2003). Actigraphy in the assessment of insomnia. Sleep, 26, 902–906. 10.1093/sleep/26.7.902 [DOI] [PubMed] [Google Scholar]
  26. Van Someren EJW (2007). Improving actigraphic sleep estimates in insomnia and dementia: How many nights? Journal of Sleep Research, 16, 269–275. 10.1111/j.1365-2869.2007.00592.x [DOI] [PubMed] [Google Scholar]
  27. Weiss AR, Johnson NL, Berger NA, & Redline S (2010). Validity of activity-based devices to estimate sleep. Journal of Clinical Sleep Medicine, 6, 336–342. [PMC free article] [PubMed] [Google Scholar]

RESOURCES