The Utility of Actigraphy to Measure Sleep in Chronic Pain Patients and Its Concordance with Other Sleep Measures: A Systematic Review and Meta-Analysis

There may be a bidirectional relationship between sleep and pain in patients with chronic pain. Actigraphy is increasingly being used as a non-invasive and objective method to assess sleep in chronic pain patients. This systematic review aimed to evaluate the utility of actigraphy in chronic pain patients. Additionally, meta-analyses were conducted to compare sleep parameters measured by actigraphy with those measured by sleep diary and polysomnography. Medline (1946-2019), Medline In-Process (May 2019), Embase (1947-2019), Cochrane Central Register of Controlled Trials (1991-2019), Cochrane Database of Systematic Reviews (2005-2019), and PubMed-NOTMedline (1946-2019) were searched for studies using actigraphy to measure sleep in chronic pain patients. Using the random effects model, meta-analyses were conducted to examine the concordance of actigraphy versus sleep diary and actigraphy versus polysomnography for commonly measured sleep parameters. Thirty-four studies with 3,590 patients were included. As an adjunct to sleep diary, actigraphy detected improvements in various sleep parameters after interventions in 10 studies and provided a useful objective sleep metric when comparing pain patients with healthy subjects in four studies; however, diary measurements were more “ sensitive ” . Comparing sleep diary versus actigraphy, sleep onset latency was significantly lower with actigraphy (mean difference of 22.7 minutes lower; 95% confidence interval: 13.2 to 32.2 minutes lower; p<0.01). No sleep parameters were significantly different between polysomnography and actigraphy; however, the confidence intervals were large. Actigraphy is an objective assessment tool that is being increasingly utilized to measure sleep in chronic pain patients. Based on studies that have measured sleep with both sleep diary and actigraphy, there are intrinsic differences between the two assessment methods as actigraphy lacks the cognitive component of subjective measures. Even though no differences in sleep parameters were detected between actigraphy and polysomnography, it cannot be established that the two are equivalent objective measures because of the limited number of studies and large variability.


INTRODUCTION
It is estimated that up to 50-80% of chronic pain patients report sleep disturbances [1][2][3]. The relationship between pain and poor sleep is not fully elucidated and likely bidirectional [4]. Limited research suggests that sleep has a greater impact on pain than pain on sleep [2,5]. It is important to assess sleep problems in chronic pain patients to treat their sleep and pain. The gold standard for sleep assessment is polysomnography (PSG); however, PSG is expensive and requires many physiological monitors [6]. A sleep diary is another tool that can measure sleep for longer periods, but they are subjective and can be cumbersome to complete [7].
Actigraphy is a non-invasive and objective method to assess sleep. It uses a small actigraph monitor, usually a wristwatch-like device that contains an accelerometer to measure motor activity over predefined periods (epochs) [8]. The activity in each epoch is then analyzed by computer software and defined as either "sleep" or "wake" [8]. Compared to PSG, actigraphy presents several advantages: 1) it allows for continuous monitoring from days to weeks; 2) it allows for monitoring in the patient's normal sleep environment; 3) it is less invasive (i.e. patient only has to wear watch versus being attached to several monitors); and 4) it is of lower cost [8]. The Clinical Practice Guideline on Actigraphy developed by the American Academy of Sleep Medicine suggests using actigraphy as an adjunct for assessing sleep in insomnia and circadian rhythm sleep-wake disorder [9].
Sleep parameters that are commonly assessed by actigraphy, sleep diary, and PSG include total sleep time (TST; total duration of sleep during the major sleep period), sleep efficiency (SE; proportion of time the patient is asleep during the total time in bed), sleep onset latency (SOL; duration between getting in bed and falling asleep), and wake after sleep onset (WASO; duration of time awake after initial sleep onset and before getting out of bed) [10]. Although correlated, discrepancies exist between the different assessment methods [7]. In particular, these discrepancies have been demonstrated to be larger in certain patient populations, such as patients with insomnia and chronic conditions [11,12].
Actigraphy is a potentially promising tool to diagnose and treat sleep disorders in chronic pain patient populations; however, the use of actigraphy in this population has not been systemically examined. Additionally, the discrepancy between actigraphy versus sleep diary or PSG has not previously been evaluated. The primary objective of this systematic review is to qualitatively assess the utility of actigraphy in the following contexts: 1) evaluating sleep after an intervention, 2) investigating the relationship between sleep and pain, 3) assessing the relationship between sleep and physical activity, and 4) comparing sleep in pain patients with healthy subjects. The secondary objective is to conduct a meta-analysis to determine the discrepancy between actigraphy versus sleep diary and actigraphy versus PSG in chronic pain patients.

Inclusion and exclusion criteria
Studies were included if they met the following criteria: 1) the participants were chronic pain patients (experiencing pain for ≥ 3 months) [13]; 2) actigraphy was used to measure sleep parameters; 3) the sleep measurement period was ≥ 5 days; 4) the participants were adults (>18 years old); 5) the studies included ≥ 15 participants (to exclude case studies and very small studies); and 6) the study was published in English. As this review is focused on the general chronic pain experience, the following types of pain conditions were excluded: 1) cancer; 2) spinal cord injuries; 3) traumatic brain injuries; 4) palliative conditions; 5) dysmenorrhea; 6) irritable bowel syndrome; 7) episodic headache, and 8) failed back surgery syndrome.

Study selection
Titles and abstracts were independently screened by two authors (DA, JS). Following the selection of abstracts, the full texts of articles identified for possible inclusion were obtained and assessed for inclusion independently by two authors (DA, JS). Disagreements were resolved by consensus or by consulting a third author (FC). performed with Review Manager 5.3 software using the random effects model [14]. For quality assessment, observational studies were assessed using the Newcastle-Ottawa Scale (which was adapted for cross-sectional studies) [15,16] and randomized controlled trials (RCTs) were assessed using the Cochrane Risk of Bias Tool [17].

Search results
The search yielded 2,060 results, with 1,252 studies remaining after duplicates were removed. Screening of titles and abstracts provided 77 studies for full text review. After full text review, 34 studies with 3,590 patients were included for analysis ( Figure 1) .

Study characteristics
Twenty-five studies were observational studies, and nine studies were RCTs. Study locations included the United States (20), United Kingdom (5), Ireland (2), Spain (2), Norway (1), Australia (1), and Brazil (2). Ten studies were on fibromyalgia, eight on arthritis, and five on chronic back pain, two on chronic migraine, and nine on chronic pain patients in general. The study characteristics and findings are presented in Table 1. Actigraph settings are presented in Table 2. Quality assessment of studies is presented in Table 3. Apart from two studies with poor quality [27,33], the remaining studies ranged from fair to good quality.     The Utility of Actigraphy We examined the utility of actigraphy to: 1) evaluate sleep after an intervention, 2) investigate the relationship between sleep and pain, 3) assess the relationship between sleep and physical activity, and 4) compare sleep in pain patients with healthy subjects.

Evaluating sleep after an intervention:
The most common intervention examined was cognitive behavioral therapy (CBT) for insomnia, which was assessed in seven studies with 715 patients (Table 1) [18][19][20][21][22][23][24]. Actigraphy was able to detect postintervention improvements in various sleep parameters in six studies. Additionally, these studies showed improvements in subjective measures of sleep with CBT [18][19][20][21][22][23]. However, one study only showed improvements in subjective measures, while not finding any differences in actigraphic parameters between the CBT and control groups [24]. Two studies that also measured sleep with PSG found that the PSG findings agreed with the actigraphic findings [20,24]. Various other interventions were also assessed, and actigraphy was able to detect improvements in select sleep parameters [25][26][27][28].

Assessing the relationship between sleep and physical activity:
Two studies with 151 patients examined the effect of sleep on physical activity and found that improved sleep led to greater physical activity levels during the day (Table 1) [37,38]. However, one of the studies only found the relationship with subjective sleep quality and not actigraphic sleep parameters [38]. Another study utilized actigraphy to examine the effect of physical activity on sleep, finding that greater fluctuations in daytime physical activity levels predicted reduced sleep duration and greater average daytime physical activity predicted greater time awake at night [39].

Comparing sleep in pain patients versus healthy subjects:
Five studies with 356 patients used actigraphy to compare sleep in chronic pain patients versus healthy subjects [40][41][42][43][44], all studies except for one [44] found that sleep was worse in pain patients (Table 1). One study showed that chronic pain patients had greater time in bed than healthy subjects; however, sleep diary was also able to detect that chronic pain patients had higher SOL and WASO [40]. Fibromyalgia patients had delayed sleep onset, delayed waking, increased time in bed, and increased WASO compared to healthy subjects [41,42]. Chronic back pain patients had increased time in bed, TST, and SOL compared to healthy subjects; in contrast, PSG measurements did not detect a difference [43]. Interestingly, in a small study (n=40), chronic migraine patients did not have any differences in actigraphic sleep parameters compared to healthy subjects [44].

Meta-analysis
The concordance between objective and subjective measures of sleep was studied in five studies in fibromyalgia patients with mixed findings [42,[45][46][47]50]. To further explore the concordance between different sleep assessment methods, metaanalyses were performed to compare actigraphy versus diary ( Figure 2) and actigraphy versus PSG ( Figure 3) for TST, SE, SOL, and WASO. The only significant difference was between actigraphy and diary in the measurement of SOL. Only four studies compared actigraphy with PSG whereas 12 studies compared actigraphy with sleep diaries.

Feasibility of using actigraphy
One third of the 34 studies reported compliance with actigraphy. Compliance was high with six studies reporting >90% of participants completing actigraphy monitoring [25,29,31,36,38,47], and the other studies reporting between 42-77% of participants [27,28,35,37]. The rate of device malfunction was low, between 4% and 6.7% [25,27]. Regarding patient satisfaction, one study implemented a questionnaire and found that most patients were satisfied; 85% of participants said they would wear an actigraph again, and 78% said it was a userfriendly method of measuring sleep [27]. A second study also reported that patients found the actigraph easy to apply and use [18].

DISCUSSION
There is a growing interest in the use of actigraphy to measure sleep in chronic pain patients. This review aimed to provide an overview of the utility of actigraphy to measure sleep in chronic pain patients and to quantitatively summarize the concordance of actigraphy with sleep diary and PSG in this patient population. Actigraphy has been used to measure sleep in several contexts, providing a useful objective measurement; however, there are intrinsic differences between actigraphy and subjective measures as actigraphy lacks the cognitive component of subjective measures. Additionally, actigraphy estimates a significantly lower SOL compared to sleep diary. While no differences in any parameters were detected between actigraphy and PSG, the number of studies was limited and the variability was large.
An D, et al.

Evaluating sleep after an intervention
After interventions designed to improve sleep, the postintervention improvement in actigraphic sleep parameters varied between studies [18][19][20][21][22][23]. In general, sleep diaries detected changes in more of the measured sleep parameters compared to actigraphy [19,22]. Additionally, one study found improvements with diary but not with actigraphy after the CBT intervention [24]. It appears that subjective measures of sleep are more "sensitive" than actigraphy. This is reasonable especially in the context of CBT for insomnia, which is a sleep disorder diagnosed based on self-reported symptoms [52]. Additionally, two studies showed that actigraphy agreed with PSG findings, providing some evidence that the two objective methods are consistent [20,24]. As such, actigraphy is able to provide an objective assessment along with other subjective measures when assessing sleep as an outcome of an intervention.

Investigating the relationship between sleep and pain
While some studies found an association between actigraphic sleep parameters and pain (29)(30)(31), other studies found no association [31][32][33][34][35][36]. Interestingly, the studies reporting an association included patients with comorbid insomnia, which could be a confounder. In studies that found no association between actigraphic sleep parameters and pain, there was often an association between diary sleep parameters and pain [34,36]. It appears that subjective measures of sleep are more "sensitive" to detect relationships between sleep and pain than actigraphy. Most prior studies that have found relationships between sleep and pain utilized subjective measures of sleep [2]. The degree of functional impairment caused by chronic pain for different individuals has an important cognitive component, comprised of pain-related beliefs and tendency to catastrophize [53]. There may be a cognitive component mediating the relationship that self-report measures are better able to detect. The utility of actigraphy to investigate the relationship between sleep and pain is inconclusive and should be further investigated.

Investigating the relationship between sleep and physical activity
The major advantage of actigraphy is that it can objectively measure both sleep and physical activity levels. Physical activity is an important outcome to assess because it is effective at preventing and reducing chronic pain [54,55]. However, we found limited studies assessing the relationship between sleep and physical activity using actigraphy in chronic pain patients, necessitating future research in on this topic.

Comparing sleep in pain patients versus healthy subjects
Consistent with prior studies, actigraphy also demonstrated that chronic pain patients had worse sleep compared to healthy subjects [2,56,57]. One small study (n=40) of chronic migraine patients did not find a difference in actigraphic sleep parameters between patients and healthy subjects. A previous study showed that migraine patients have worse subjective sleep quality [58]. Larger studies using actigraphy should be conducted to quantify sleep disturbances in chronic migraine patients. In general, actigraphy appears to be a useful tool to quantify sleep disturbances in chronic pain patients.

Concordance between actigraphy with diary and PSG
Comparing subjective measures of sleep versus actigraphy, SOL was significantly less with actigraphy while the other parameters were not significantly different. In chronic pain patients, insomnia is a common comorbidity [59,60]. Insomnia patients often report higher SOL than actigraphy; these patients possibly have a tendency to lie awake in bed without moving, which actigraphy detects as "sleep" [7,61]. Several studies in our metaanalysis also specifically recruited patients with comorbid insomnia [20,22,42,46,49]. As such, this could be a contributing factor to greater self-reported SOL amongst chronic pain patients. Previous studies have also reported that higher pain ratings in chronic pain patients are associated with longer subjective measures of SOL [62,63], suggesting a connection of subjective SOL with pain-related cognition.
Comparing PSG versus actigraphy, all four sleep parameters were comparable with no significant differences. Although not significant, TST was greater with actigraphy versus PSG and WASO was less with actigraphy versus PSG. The trend of actigraphy to overestimate " sleep " (TST) and underestimate "wake" (WASO) has been previously reported in a review that compared actigraphy versus PSG in patients with chronic conditions [12]. In patients with chronic pain or other chronic conditions, there could be changes to their mobility. This could alter the ability of actigraphy to accurately assess sleep because most actigraphy analysis software use algorithms that are only validated against healthy populations. One actigraph device, Actigraph-Insomnia, has a feature that aims to improve estimates of "wake" [38]. It involves a pressure sensor that patients hold between their finger and thumb, which is released when patients fall asleep and their muscle tone relaxes. The device could be tested in the future to see if it can improve the discrepancy. Based on thresholds set by the 2018 American Academy of Sleep Medicine on actigraphy [7], the 95% CI of the mean differences in our study were large and suggests that the two methods (actigraphy and PSG) cannot be used interchangeably. As such, even though no significant differences were found, it does not necessarily suggest that the two measurement methods are consistent and produce the same measurements.

LIMITATIONS
This review has some limitations. With the restriction to English-language studies, we may not have identified all relevant literature. The included studies were heterogeneous with regards to patient populations and study designs. This variability likely led to the high statistical heterogeneity in the meta-analysis findings. Also, our meta-analyses comparing actigraphy with PSG contained few and small studies (total n=296). As well, five of the 12 studies comparing actigraphy with diary had less than 50 subjects.

CONCLUSION
Actigraphy is an objective assessment tool that is being increasingly utilized to measure sleep in chronic pain patients. As an adjunct to subjective measures like sleep diary, actigraphy provides a useful objective sleep metric when assessing the effect of an intervention and when comparing pain patients with healthy subjects; however, there are intrinsic differences between the assessment methods, and it is unclear which measure is more suitable for these uses. Diary measurements also tend to overestimate SOL compared to actigraphy. Even though no difference was detected between actigraphic and PSG parameters, it cannot be established that the two are equivalent measures due to limited studies and large variability in values. As actigraphy presents many potential advantages, further research is needed to compare the different assessment methods with large RCTs measuring sleep using multiple assessment methods in chronic pain patients.

ACKNOWLEDGMENTS
DA devised the protocol, performed title/abstract and full-text screening, extracted outcomes, created Tables and Figures, and wrote and edited the manuscript. JS performed title/abstract and full-text screening, extracted outcomes, and wrote and edited the manuscript. JW wrote and edited the manuscript. CS wrote and edited the manuscript. SM wrote and edited the manuscript. ME conducted the literature search and edited the manuscript. FC devised the protocol, and wrote and edited the manuscript.

CONFLICT OF INTEREST
JW reports grants from the Ontario Ministry of Health and Long-Term Care, Anesthesia Patient Safety Foundation, Acacia Pharma, Merck Inc. outside of the submitted work. JW is supported by a Merit Research Award from the Department of Anesthesia, University of Toronto. FC reports research support from the Ontario Ministry of Health and Long-Term Care, University Health Network Foundation, Acacia Pharma, Medtronics grants to institution outside of the submitted work, and previous research grants from Pfizer. Up-to-date royalties, STOP-Bang proprietary to University Health Network. The other authors have no conflicts of interest to report.