Acta Psychiatr Scand 2014: 1–11 All rights reserved DOI: 10.1111/acps.12367

© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd ACTA PSYCHIATRICA SCANDINAVICA

Meta-analysis

Sleep in patients with remitted bipolar disorders: a meta-analysis of actigraphy studies Geoffroy PA, Scott J, Boudebesse C, Lajnef M, Henry C, Leboyer M, Bellivier F, Etain B. Sleep in patients with remitted bipolar disorders: a meta-analysis of actigraphy studies. Objective: Sleep dysregulation is highly prevalent in bipolar disorders (BDs), with previous actigraphic studies demonstrating sleep abnormalities during depressive, manic, and interepisode periods. We undertook a meta-analysis of published actigraphy studies to identify whether any abnormalities in the reported sleep profiles of remitted BD cases differ from controls. Method: A systematic review identified independent studies that were eligible for inclusion in a random effects meta-analysis. Effect sizes for actigraphy parameters were expressed as standardized mean differences (SMD) with 95% confidence intervals (95% CI). Results: Nine of 248 identified studies met eligibility criteria. Compared with controls (N = 210), remitted BD cases (N = 202) showed significant differences in SMD for sleep latency (0.51 [0.28–0.73]), sleep duration (0.57 [0.30–0.84]), wake after sleep onset (WASO) (0.28 [0.06– 0.50]) and sleep efficiency ( 0.38 [ 0.70–0.07]). Moderate heterogeneity was identified for sleep duration (I2 = 44%) and sleep efficiency (I2 = 44%). Post hoc meta-regression analyses demonstrated that larger SMD for sleep duration were identified for studies with a greater age difference between BD cases and controls (b = 0.22; P = 0.03) and nonsignificantly lower levels of residual depressive symptoms in BD cases (b = 0.13; P = 0.07). Conclusion: This meta-analysis of sleep in remitted bipolar disorder highlights disturbances in several sleep parameters. Future actigraphy studies should pay attention to age matching and levels of residual depressive symptoms.

P. A. Geoffroy1,2,3,4,5, J. Scott6,7,

C. Boudebesse5,8,9, M. Lajnef9, C. Henry5,8,9,10, M. Leboyer5,8,9,10, F. Bellivier1,2,3,4,5, B. Etain5,8,9,10 1

Inserm, UMR-S 1144, Paris, France, 2AP-HP, GH SaintLouis - Lariboisi
ere - Fernand Widal, P^ole Neurosciences, Paris Cedex 10, 3Universite Paris Descartes, UMR-S 1144, Paris, 4Universite Paris Diderot, UMR-S 1144, Paris, 5Fondation FondaMental, Creteil, France, 6Academic Psychiatry, Institute of Neuroscience, Newcastle University, Newcastle Upon Tyne, 7Centre for Affective Disorders, Institute of Psychiatry, London, UK, 8 AP-HP, H^opital H. Mondor - A. Chenevier, DHU PePsy, P^ole de Pstychiatrie, 9INSERM, U955, Equipe Psychiatrie Genetique, and 10Universite Paris Est, Faculte de medecine, Creteil, France

Key words: bipolar disorders; actigraphy; euthymia; sleep; circadian rhythms Pierre Alexis Geoffroy, Service de Psychiatrie Adulte (Pr Bellivier), H^opital Fernand Widal, 200, rue du Faubourg Saint-Denis, 75475 Paris Cedex 10, France. E-mail: [email protected]

Accepted for publication October 21, 2014

Summations

• This meta-analysis demonstrates that, compared with non-psychiatric controls, cases with bipolar • •

disorders experience a range of sleep abnormalities during remission. In clinical practice, systematic assessment of sleep quality in bipolar disorders is indicated even during remission, including monitoring of sleep latency, duration, efficiency and wake after sleep onset (WASO) using actigraphy when available or a subjective quality sleep measure such as the PSQI that incorporates these parameters in the generated subscores. Methodological improvements for future actigraphy studies in bipolar disorder cases in remission should particularly include more careful matching for age and reliable assessment of any residual symptomatology.

1

Geoffroy et al.

Considerations

• Several potential confounding factors were not reported in all studies, so we were able only to assess • •

a limited number of sources of heterogeneity (e.g. incomplete matching for age, sample size, level of mood symptoms). We restricted our analysis to widely reported, standard sleep parameters, but other measures such as sleep–wake and activity cycles or variability (rather than mean scores) in sleep markers may be more sensitive to discriminate patients from controls. Nearly, all BD cases were taking psychotropic medication, whilst nearly all controls were medication free. However, it was not possible to determine to what extent the treatment of BD may influence the case–control differences in sleep measures.

Introduction

Bipolar disorders (BDs) are severe, chronic mental disorders affecting 1–4% of the population worldwide (1), and they are ranked among the most burdensome diseases globally (2). BDs are characterized by alternating phases of major depressive episodes and hypomanic or manic episodes with intermittent periods of remission (when individuals have few or no symptoms of BDs for three or more months) (3). Although the pathophysiological determinants of BDs remain unknown, it is well demonstrated that genetic and environmental vulnerabilities are both important (4). In addition, there is increasing evidence that abnormalities of circadian rhythmicity and sleep homeostasis may differentiate BD cases from healthy controls (5). Sleep and circadian abnormalities were first considered as markers of the acute phases of BD, with the classic patterns being the presence of insomnia or hypersomnia during major depressive episodes and a decreased need for sleep without subsequent fatigue in (hypo)manic episodes (6, 7). It is now recognized that sleep abnormalities may also occur and/or persist during remission (6–10). Alterations of sleep quality and continuity during interepisode phases of BDs are prevalent. For example, 83% patients with BDs in remission reported poor sleep quality compared to 21% of healthy controls (11) and Harvey et al. observed that 55% BD cases also met diagnostic criteria for insomnia (10). The identification of sleep and circadian biomarkers during remission is of crucial importance as disruptions in the sleep homeostasis and circadian rhythms irregularities are frequently associated with relapses. For example, in a sample of euthymic patients with BDs, the presence of sleep disturbance was significantly associated with greater risk for mood episode recurrence (12), whilst hypersomnia in the interepisode period is associated 2

with future depressive symptoms (13). Furthermore, a systematic review of the prodromes of manic and depressive episodes identified the importance of sleep as an early symptom of acute relapses of either polarity (14). Taken together, these findings emphasize the importance of monitoring sleep in BD cases, as sleep disturbance appears to be a symptom of a current mood episode, a possible trigger of future relapses and a prodromal marker of an emerging mood episode. Several tools have been used to explore sleep and circadian abnormalities, but actigraphy is an increasingly popular, non-invasive, objective measure of sleep/wake irregularities in affective and other psychiatric disorders (15). Actigraphy uses a mobile portable device called an Actiwatch that comprises an accelerometer to record the amplitude and frequency of movements which can be used as markers of a individual’s sleep/wake cycle (15). In patients with psychiatric disorders, actigraphy presents several advantages compared to the gold standard assessment, namely polysomnography (PSG). For example, actigraphy (i) allows recordings across naturalistic conditions, (ii) can record data over a longer duration than PSG, (iii) is easier to use and to access in clinical settings, (iv) is less expensive and (v) is minimally disruptive. However actigraphy, compared with PSG, does not allow examination of sleep architecture and may under- or overreport some aspects of the sleep–wake cycle or sleep-related events (16, 17). Despite its well documented and widespread use in chronobiological studies, actigraphy has only been applied in BD research consistently for the last decade (18). As a number of actigraphy studies of BD cases have demonstrated abnormalities in sleep quantity and quality compared to controls, there is a need for a meta-analysis on sleep in remitted BD cases focused on actigraphy studies. Such information may be particularly useful in identifying markers

Actigraphy meta-analysis in bipolar disorders of the disease process that persist during remission and which may therefore represent an important element of the underlying pathophysiology. However, we also anticipated that different methodologies (e.g. the presence of residual mood symptoms; selection of controls) might also influence any apparent between group differences. Aims of the study

We undertook a systematic review and a metaanalysis of published actigraphy studies with the dual aims of identifying the following: i Sleep parameters that consistently differentiate bipolar disorder cases in remission from nonpsychiatric controls and ii Potential sources of heterogeneity in these case– control studies. Material and methods

We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) recommendations to undertake the search and analysis (19); Fig. 1 shows the flow diagram for the study. Eligibility

Identification

English language publications reporting a case– control actigraphy study that recorded sleep data

for at least 24 h in adult participants (age ≥18 years) were eligible for inclusion. For BD cases, we required an established clinical diagnosis of BD I, II or BD-NOS subtypes (with information on the proportion of cases in each subgroup) and a clear definition of the criteria used to define remission. For control participants, we required a statement in the methodology that they had no known DSM-IV Axis I psychiatric disorder(s). Studies were excluded if there was insufficient information on the actigraphy methodology (methods and duration of recording, inadequate data on sleep parameters), if the details or data were not available from the original researchers. Also, only one set of findings were included from each independent data set identified (i.e. if two publications reported from the same data set, the most relevant publication was selected and/or the research group was contacted to identify the most appropriate data to include in the analysis). Then, we defined exclusion criteria for the eligible studies to be retained in the meta-analyses, which were the following: (i) if the study lacks from relevant data (e.g. patients with acute mood symptoms or samples including patients with unipolar disorders); (ii) if the study did not include any information on current manic or depressive symptoms; (iii) if participants of the study were outside the prerequisite age range (children or adolescent cases); and (iv) when patients or controls were diagnosed as having sleep disorders such as

Records identified through database searching a (n = 232)

Additional records identified through other sources (n = 23)

Fig. 1. Flow Diagram of Actigraphy Studies’ Selection Process. aThe literature search was performed from the PubMed electronic database and using the Mesh heading: bipolar disorder AND (actimetry OR actigraphy OR actigraph* OR actimet* OR accelerometer OR sleep latency OR sleep efficiency OR wake after sleep onset OR sleep duration).

Included

Eligibility

Screening

Records after duplicates removed (n = 248)

Records screened (n = 248)

Studies assessed for eligibility (n = 31)

Studies included in meta-analysis (n = 9) Total of 412 cases

Records excluded (n = 217)

Studies excluded (n = 22), if: -Patients were symptomatic or with unipolar disorder: 18 -No information of current manic symptoms: 3 - Cases were children or adolescents: 1 - Patients with a diagnosis of a sleep disorders: 0

3

Geoffroy et al. sleep apnoea syndrome, restless leg syndrome, narcolepsy, and Kleine–Levin syndrome. Search strategy

We searched PubMed and Embase databases for publications between January 1985 and 1 June 2014. The following MESH terms were used: bipolar disorder AND (actimetry OR actigraphy OR actigraph* OR actimet* OR accelerometer OR sleep latency OR sleep efficiency OR wake after sleep onset OR sleep duration). We also examined the citation lists of identified publications for additional studies, used the related articles function of the PubMed database and searched Google Scholar for other relevant sources of data. Study selection

We first excluded duplicate publications. Three authors (BE, PAG and CB) independently screened the titles of potentially eligible publications. Some papers were excluded at this stage (see Fig. 1), but the abstracts of potential inclusions were then reviewed jointly by this group. After review of abstracts and papers, likely inclusions were assessed by two authors (PAG and BE) who independently extracted all data regarding samples characteristics (recruitment strategy, mean age and gender distribution, level of mood symptoms) and actigraphy measures (including monitoring device, software packages used, duration of recording, etc.). Any discrepancies in data extraction were corrected by re-examining the original publications and consensus agreement. When data were missing, researchers were contacted to for further information. Choice of sleep parameters and potential confounders

A priori, we determined that any sleep parameter that was reported by two or more actigraphy studies was eligible for inclusion in the meta-analysis. Four sleep parameters were reported by most studies: sleep latency, sleep efficiency, sleep duration and WASO (wake after sleep onset). There were insufficient data for meta-analysis for sleep fragmentation index, amplitude, stability, variability or activity-related markers. We identified a priori that age, symptom levels, sample size or duration of recording may be a potential confounder. For example, we assumed that sleep patterns change with age, that short duration of actigraphy recording may under- or overestimate between group differences and/or that residual depressive symptoms may affect acti4

graphic parameters (20, 21). However, we also used a simple criterion that –to assess a variable or factor in a meta-regression – it has to be reported in more than 30% of the studies we identified. As such, regression analyses were possible for duration of actigraphy recording (in days), age matching (the absolute value of the difference between the mean age at assessment of cases compared with controls), total sample size and residual mood symptoms (HDRS – Hamilton Depression Rating Scale – and YMRS – Young Mania Rating Scale – scores). For residual depressive symptoms, we used a ‘conversion’ table provided on the Inventory for Depressive Symptoms website (www.ids-qids.org) which enables scores on a range of symptom scales (in our study, MADRS: Montgomery-Asberg Depression Rating Scale and IDS-C: Inventory for Depressive Symptoms Clinician version) to be transformed into an HDRS equivalent score. Data analysis

We extracted means and standard deviations (SD) for the four actigraphic sleep markers identified and undertook a separate meta-analysis for each variable using Review Manager 5.2 (Cochrane Collaboration, Copenhagen, Denmark). Random effects modelling for pooled effect sizes (ES) was used because it provides a more conservative ES estimate (22). We calculated the standardized mean difference (SMD) and 95% confidence intervals (95% CI) for each study, with the SMD defined as the difference in means between the two groups (BD cases in remission and controls) divided by the pooled standard deviation of the measurements. The SMDs were interpreted in a similar manner to Cohen’s d (0.20 = small ES; 0.50 = medium ES; 0.80 = large ES). We used Review Manager to construct forest plots of all comparisons, whilst publication bias was assessed by visual inspection of the funnel plots (Figure S1). The I2 statistic was used to quantify heterogeneity, with the values of 25%, 50% and 75% reflecting a small, moderate or high degree of heterogeneity respectively (23). Meta-regression of case–control analyses was performed using the METAREG function of STATA software 12, which performs random effects meta-regression using aggregate-level data (24). This function uses an iterative method to produce estimates (b (SD)), P-values and graphs for the outputs (24). For meta-regressions, random effects modelling carries a least restrictive set of assumptions, which is particularly relevant in an area of study where significant between-study variability is to be expected. In addition, these meta-regressions

Actigraphy meta-analysis in bipolar disorders relate to potential confounders that were selected post hoc, according to their frequency of reporting in the studies meeting eligibility criteria for the meta-analyses.

Results Search results

As shown in Fig. 1, the initial search returned 248 records (after the removal of duplicates). Following preliminary screening of the 248 titles and abstracts, 217 records were excluded, whilst 31 were reviewed in detail. The 22 excluded studies comprised of 18 studies that lacked relevant data (patients with acute mood symptoms or samples including patients with unipolar disorders) (20, 25–41), three studies that did not include any information on current manic symptoms (42–44) and one study whose participants were outside the prerequisite age range (children or adolescent cases) (45). Therefore, nine independent data sets were included in the meta-analysis (10, 18, 46–52). The nine studies had a median sample size of 40 participants (range 27–68) and included a total of 412 individuals (202 BD cases in remission and 210 controls). The majority of cases had BD I, but many samples also included BD II or BD-NOS cases. Controls were predominantly individuals with no evidence of a mental disorder according to DSM-IV criteria, although a few studies used other criteria to define the non-psychiatric group (see Table 1). The methodologies and measurements also varied considerably in terms of levels of residual depressive and (hypo)manic symptoms at inclusion, degree of matching for age and gender, and the definition of ‘controls’ (unscreened controls, screened controls, ‘good sleepers’). All studies asked participants to wear an actigraph on the wrist of their non-dominant hand, but various actigraphy devices and software packages were used for the analyses (at least seven different kinds). The median duration of actigraphic recording was 7 days (range 3–54 days); variable sampling algorithms were employed (with a median epoch of 60 s), and a range of sleep parameters was extracted (see Table 1 and Table S1). Meta-analysis

Separate meta-analyses were undertaken for each of the four sleep parameters that were eligible for analysis (see Fig. 2a–d). Sleep Latency (7 studies): BD cases presented with longer sleep latency with the SMD between cases and controls of 0.51 [0.28–0.73] (z = 4.43

P < 0.00001). Meta-analysis showed no heterogeneity between studies (I2 = 0%). See Fig. 2a. Sleep Duration (9 studies): BD cases had longer sleep duration with a SMD between cases and controls of 0.57 [0.30–0.84] (z = 4.16 P < 0.0001). Meta-analysis showed moderate homogeneity between studies (I2 = 44%). See Fig. 2b. Wake After Sleep Onset (WASO) (7 studies): The SMD for WASO between BD cases and controls was 0.28 [0.06–0.50] (z = 2.48 P = 0.01). Meta-analysis showed no heterogeneity between studies (I2 = 0%). See Fig. 2c. Sleep Efficiency (7 studies): BD cases had lower sleep efficiency than controls, and the SMD was 0.38 [ 0.70–0.07] (z = 2.40 P = 0.02). Metaanalysis showed moderate homogeneity between studies (I2 = 44%). See Fig. 2d. Publications bias

Visual examination of funnel plots suggested no publication biases for sleep duration, sleep latency and WASO parameters (see Figure S1). However, the sleep efficiency plot demonstrated one outlying study (52). Meta-regressions

For sleep duration, larger SMD were reported for studies with greater age differences between BD cases and controls (b = 0.22; P = 0.03) and trended towards statistical significance for depressive symptoms, with lower level of depressive symptoms associated with longer sleep duration (b = 0.13; P = 0.07). The SMD did not appear sensitive to the duration of actigraphic recording or total sample size (see Table 2). For sleep efficiency, SMD appeared sensitive to none of the assessed potential confounders (see Table 2 and additional Figure S2). Discussion

This is among the first meta-analysis that demonstrates that, compared with control participants, BD cases in remission have longer sleep latency, sleep duration and WASO, and lower sleep efficiency. This suggests that BD cases experience pervasive sleep abnormalities during remission phases (i.e. when no or minimal other illness symptoms are present). Research has increasingly focussed on sleep/circadian dysregulation in remitted patients with BDs, and this meta-analysis of actigraphy provides a timely overview of the state of the evidence. This meta-analysis helps to delineate potentially reliable interepisode sleep biomarkers of 5

6

32

14

54

14

21

7

St-Amand et al. (2013) (50) (Canada)

Geoffroy et al. (2014) (51) (France)

McKenna et al. (2014) (52) (USA)

BDI

16 BDI 8 BDII 2 BDNOS

11 BDI 3 BDII

BDI

19 BDI 3 BDII

BDI

BDI

BDI

BDI

Diagnosis

SCID-IV

DIGS (DSM-IV-TR)

SCID

SCID

SCID-I SCID-II

SCID-IV

SCID

DSM-IV case note information SCID

Assessment interview

HDRS and YMRS

MADRS and YMRS

BDI-II, HDRS and YMRS

IDS-Cand YMRS

HDRS and YMRS

HDRS and YMRS

Did not meet DSM episode criteria HDRS and YMRS HDRS and MAS

Remission criteria Self-reported

YMRS = 2.3 (2.8) MAS = 4.7 (3.8) YMRS = 3.4 (5.0)

YMRS < 10 YMRS = 3.2 (3.0)

YMRS = 1.4 (1.3)

YMRS = 0.65 (1.38)

YMRS = 1.21 (1.34)

HDRS = 1.4 (1.9) HDRS = 8.5 (4.8) HDRS = 4.6 (4.8)

HDRS-17 < 25 IDS-C = 8.6 (4.7)

HDRS = 5.4 (4.3)

MADRS = 1.85 (2.80)

HDRS = 3.57 (2.36)

Manic symptom rating

Self-reported

Depression symptom rating

49.07 (11.34) [NA]

53.50 (11.49) [32–74]

44.6 (11.0) [23–66]

34.7 (10.5) [18–64]

32.7 (10.0) [NA]

44.4 (9.8) [NA]

39.6 (15.2) [NA] 44.37 (13.10) [NA]

47.3 (10.61) [26–68]

Mean age (SD) [range]

79%

65%

50%

63%

41%

81%

74%

50%

58%

% Female

NA

96%

100%

94%

100%

100%

100%

100%

95%

% treated†

14

29

13

36

28

32

19

20

19

N

Unaffected controls†† of similar age and sex distribution Unaffected‡ Gender, age and education comparable

No axis I disorder and normal sleep¶ No insomnia or mental disorder

Unaffected‡ Gender and age matched Unaffected‡ Similar age and gender distribution Unaffected‡

Unaffected‡ Gender and age matched Good sleeper§

Characteristics

SCID-IV

SCID and Insomnia interview schedule DIGS (DSM-IV-TR)

SCID

SCID-I SCID-II

SCID-IV

SCID

SCID

SCID-I

Assessment interview

Controls

46.36 (15.04)

54.10 (9.11)

47.15 (10.4)

33.3 (12.6)

28.3 (7.2)

42.3 (10.8)

46.89 (14.82)

35.0 (13.4)

45.8 (10.93)

Mean age (SD)

71%

45%

46%

53%

43%

75%

74%

65%

58%

% Female

BD, bipolar disorder (I, II or NOS subtype); BDI-II, Beck Depression Inventory-II; HDRS, Hamilton Depression Rating Scale; IDS-C, Inventory of Depressive Symptomatology-Clinician version; MAS, Bech–Rafaelson Mania Assessment Scale; NA, not available; P-YMRS, parent version–Young Mania Rating Scale; SCID-I, Structured Clinical Interview for DSM-IV, Axis-I Disorders; SD, standard deviation; UK, United Kingdom; YMRS, Young Mania Rating Scale. †Percentage of patients with BD and treated with one or more psychotropic medications. ‡Unaffected = means of clinical assessment did not reveal any evidence of mental disorder. §Indication of sleeping ‘very well’ on the Insomnia Diagnostic Interview, no difficulties with sleep in the past month, and no current use of medication for sleep. ¶Normal sleep was defined as the absence of any diagnosable sleep disorder (including insomnia) based on the Duke Structured Clinical Interview for Sleep Disorders (Edinger et al., 2004), and scoring below established cut-offs for sleep disturbance on the Insomnia Severity Index (ISI; score ≤7; Bastien et al., 2001) and the Pittsburgh Sleep Quality Index (PSQI; score ≤5; Buysse et al., 1989). ††Patients and controls were included if during the preceding three months, they had 1) not experienced any periods of severe sleep disruption due to somatic conditions (e.g. organic insomnia/hypersomnia, or sleep wake disorders) and/or any life event that may have altered their sleep patterns; 2) not been hospitalized or received a treatment that may disrupt sleep; and 3) not been prescribed medication or taken drugs that may disrupt sleep and not changed either the dose or type of psychotropic treatment.

14

26

22

6

Ritter et al. (2012) (49) (Germany) Gershon et al. (2012) (48) (USA)

36

3

19

7

Salvatore et al. (2008) (47) (Italy)

20

8

Harvey et al. (2005) (10) (UK) Jones et al. (2005) (46) (UK)

19

5

N

Millar et al. (2004) (18) (UK)

Study (Country)

Days of actigraphy

Cases

Table 1. Characteristics of actigraphic studies exploring remitted bipolar disorder that are included in the meta-analysis

Geoffroy et al.

Actigraphy meta-analysis in bipolar disorders (a) Sleep latency Study or Subgroup

Bipolar Control Std. Mean Difference Mean SD Total Mean SD Total Weight IV, Random, 95% CI

Millar et al. 2004 (18) Jones et al. 2005 (46) Harvey et al. 2005 (10) Gershon et al. 2012 (48) Ritter et al. 2012 (49) St-Amand et al. 2013 (50) Geoffroy et al. 2014 (51)

19.50 29.23 18.50 12.00 11.15 14.60 25.23

22.10 24.89 17.80 11.60 7.53 9.02 33.65

19 19 20 32 22 14 26

8.00 17.08 13.90 7.30 6.97 15.69 11.59

6.90 14.11 8.30 3.50 4.10 9.22 7.98

19 19 20 36 28 13 29

11.80% 12.00% 13.00% 21.50% 15.30% 8.90% 17.40%

0.69 [0.03-1.34] 0.59 [–0.06-1.24] 0.32 [–0.30-0.95] 0.56 [0.07-1.04] 0.70 [0.13-1.28] –0.12 [–0.87-0.64] 0.56 [0.02-1.10]

Total (95% CI) 152 164 100.00% Heterogeneity: Tau = 0.00; Chi = 3.82, df = 6 (P = 0.70); I = 0% Test for overall effect: Z = 4.43 (P < 0.00001)

0.51 [0.28-0.73]

(b) Sleep duration Bipolar Control Study or Subgroup Mean SD Total Mean SD Total Millar et al. 2004 (18) 434.20 91.70 19 387.50 53.00 19 Jones et al. 2005 (46) 450.76 71.77 19 446.78 30.08 19 Harvey et al. 2005 (10) 504.00 78.00 20 420.00 60.00 20 Salvatore et al. 2008 (47) 512.40 102.00 36 433.20 105.00 32 Gershon et al. 2012 (48) 379.50 62.10 32 360.80 57.50 36 Ritter et al. 2012 (49) 447.32 67.62 22 374.50 54.12 28 St-Amand et al. 2013 (50) 416.84 79.54 14 416.00 31.37 13 Geoffroy et al. 2014 (51) 475.42 64.50 26 455.83 53.98 29 McKenna et al. 2014 (52) 461.02 67.31 14 423.08 25.41 14

Std. Mean Difference Weight IV, Random, 95% CI 10.30% 0.61 [–0.04-1.26] 10.60% 0.07 [–0.57-0.71] 9.90% 1.18 [0.51-1.86] 13.90% 0.76 [0.26-1.25] 14.30% 0.31 [–0.17-0.79] 11.20% 1.19 [0.58-1.80] 8.60% 0.01 [–0.74-0.77] 12.90% 0.33 [–0.21-0.86] 8.40% 0.72 [–0.04–1.49]

Total (95% CI) 202 210 100.00% Heterogeneity: Tau = 0.07; Chi = 14.20, df = 8 (P = 0.08); I = 44% Test for overall effect: Z = 4.16 (P < 0.0001)

0.57 [0.30-0.84]

(c) Wake After Sleep Onset (WASO) Study or Subgroup Millar et al. 2004 (18) Jones et al. 2005 (46) Harvey et al. 2005 (10) Gershon et al. 2012 (48) Ritter et al. 2012 (49) St-Amand et al. 2013 (50) Geoffroy et al. 2014 (51)

Bipolar Control Mean SD Total Mean SD Total 59.00 26.00 19 49.20 17.50 19 57.86 22.58 19 51.25 20.39 19 37.50 30.20 20 29.60 24.50 20 96.90 53.50 32 85.60 27.90 36 86.98 30.33 22 75.56 23.82 28 53.73 16.61 14 53.92 20.79 13 57.88 23.17 26 52.62 27.33 29

Total (95% CI) 152 Heterogeneity: Chi = 1.10, df = 6 (P = 0.98); I = 0% Test for overall effect: Z = 2.48 (P = 0.01)

(d)

Sleep efficiency

Study or Subgroup Millar et al. 2004 (18) Jones et al. 2005 (46) Gershon et al. 2012 (48) Ritter et al. 2012 (49) St-Amand et al. 2013 (50) Geoffroy et al. 2014 (51) McKenna et al. 2014 (52)

Std. Mean Difference Weight IV, Random, 95% CI 11.90% 0.43 [–0.21-1.08] 12.10% 0.30 [–0.34-0.94] 12.70% 0.28 [–0.34-0.90] 21.60% 0.27 [–0.21-0.74] 15.50% 0.42 [–0.15-0.98] –0.01 [–0.76-0.75] 8.70% 17.50% 0.20 [–0.33-0.73]

164 100.00%

Bipolar Control Mean SD Total Mean SD Total 83.00 9.20 19 86.90 3.60 19 84.22 6.70 19 86.42 4.20 19 76.10 9.40 32 77.20 6.40 36 81.85 4.67 22 81.96 4.43 28 85.40 4.86 14 85.79 4.14 13 81.54 9.87 26 84.90 6.32 29 86.55 6.40 14 94.85 3.30 14

0.28 [0.06–0.50]

Std. Mean Difference Weight IV, Random, 95% CI 13.70% –0.55 [–1.20-0.10] 13.90% –0.39 [–1.03-0.26] 18.70% –0.14 [–0.61-0.34] 16.10% –0.02 [–0.58-0.53] 11.30% –0.08 [–0.84-0.67] 16.80% –0.40 [–0.94-0.13] 9.40% –1.58 [–2.45-–0.72]

Total (95% CI) 146 158 100.00% Heterogeneity: Tau = 0.08; Chi = 10.76, df = 6 (P = 0.10); I = 44% Test for overall effect: Z = 2.40 (P = 0.02)

–0.38 [–0.70-–0.07]

Fig. 2. Meta-analysis’s forest plots of actigraphic parameter comparison of patients with remitted bipolar disorder vs. healthy controls for a) sleep latency, b) sleep duration, c) wake after sleep onset (WASO) and d) sleep efficiency.

7

Geoffroy et al. Table 2. Meta-regressions of sleep duration and sleep efficiency actigraphic parameters Sleep duration Potential confounders Duration of recording N total Residual depressive symptoms* Residual manic symptoms† Absolute difference of mean age

N studies 9 9 7 6 9

b 0.01 0.002 0.13 0.09 0.22

Sleep efficiency SD

P values

N studies

0.008 0.01 0.058 0.16 0.08

0.28 0.84 0.07 0.60 0.03

7 7 5 4 7

b 0.008 0.01 0.10 0.21 0.04

SD

P values

0.01 0.01 0.14 0.38 0.16

0.49 0.29 0.52 0.64 0.81

N, number (or sample size); SD, standard difference. *Assessed by the HDRS (Hamilton Depression Rating Scale). †Assessed by the YMRS (Young Mania Rating Scale).

BDs, some with moderate effect sizes (ES) and only minimal confounding. A methodological strength of this meta-analysis is the focus only on BD cases in remission to reduce the potential bias introduced by current symptoms of mania or depression when examining differences in sleep duration or other sleep markers. We demonstrate that persistent abnormalities in several sleep parameters are observed among patients with remitted BDs. In clinical practice, systematic assessment of sleep patterns (as a whole) and especially of sleep latency, duration, efficiency and WASO may provide important insights into the quality and stability of remission. Thus, because of its objective, naturalistic and non-invasive properties, actigraphy might be a useful clinical tool to allow regular reviews of sleep quality in patients during remitted phases and/or to evaluate treatment efficacy, which is recommended by the American Sleep Association (53). However, if access to actigraphy is difficult, the present findings encourage monitoring with subjective approaches such as self-rated questionnaires or sleep logs (11). For example, the Pittsburgh Sleep Quality Index (PSQI) may offer some advantages for routine practice as compared to actigraphy and can provide reasonable approximations of sleep profile as well as daytime dysfunction, etc. (17). Sleep logs (or diaries) are probably the very least time-consuming tools to use in routine settings and offer an interesting snapshot of sleep disturbances. Such sleep logs can be employed to identify abnormalities in the four parameters examined here: sleep latency, duration, efficiency and WASO (16). One of the other key conclusions from this meta-analysis is that whilst there is some consistency in the findings with regard to cases vs. controls, there are some weaknesses and limitations in the available literature on actigraphy. The number of independent data sets is small, and the number of eligible studies and the sample sizes are lower 8

than reported in other meta-analyses of actigraphy in depression (16 studies, 412 cases) and ADHD (16 studies, 722 cases and 638 controls) (21, 54). Given the low number of published studies; the small, often only partially matched samples (e.g for age and gender); the variable definitions of non-psychiatric controls; and the fact that we restricted our search to English language publications, we cannot exclude study and publication biases. Although all the eligible studies excluded patients with comorbid sleep disorders (sleep apnoea syndrome, restless leg syndrome, narcolepsy, Kleine–Levin syndrome), the assessment of syndromes that are more common in BD populations, such as sleep apnoea (55), was based on screening tools (like the Berlin Questionnaire for sleep apnoea) rather than more reliable but resource-intensive approaches such as laboratory assessments. Furthermore, the mean duration of remission and the number of mood episodes may have biased, to some extent, the comparison between studies. It is also noteworthy that more cases of BD I were included compared to other bipolar spectrum disorders. This may mean that – even though the cases were in remission – they may have experienced more severe or complex illness patterns, had a longer duration of BDs and may be more vulnerable to sleep problems (14). A further, insurmountable problem in a range of case–control studies (fMRI, actigraphy, PSG, etc.) is that over 90% cases were currently receiving psychotropic medications. We also were not able to consider information about sleep medication use. None of the included samples was large enough to allow further exploration of the effects of medications on the estimation of the case–control SMD. However, one of the included studies suggested that the type and dose of psychotropic medications were unrelated to actigraphic measures in patients with remitted BDs (47). Nevertheless, lithium, valproate and antipsychotics have been shown to regulate sleep and circadian rhythms in patients with BDs

Actigraphy meta-analysis in bipolar disorders (56), so we can speculate, but not prove, that the likely impact of medications may be to improve sleep in BD cases and reduce the SMD from controls. There are some issues in how actigraphy is applied and analysed in different studies. For example, there was substantial heterogeneity in the algorithms employed, Actiwatch devices used and software employed in analyses. We were not able to assess the impact of these on the results of this meta-analysis, but this should be considered in future reviews, and there would be benefit in research groups collaborating in the use of more homogeneous approaches. We also were restricted in our analysis to focusing on the essential, but rather basic, sleep parameters, but not on other derived actigraphic parameters, in particular those regarding sleep–wake and activity (such as assessed recently in acute depression (23)) or potentially more sensitive markers such as variability (rather than mean scores) in sleep parameters in BDs. The latter were only available in a few studies (18, 48, 51), so it was not realistic to undertake multiple pooled analyses on these subsets. Lastly, whilst actigraphy has become increasingly popular as a research tool, it is not without limitations. In some studies, actigraphy might overestimate total sleep time and underestimate WASO (57–59), and it may be less reliable in populations with fragmented sleep and in measuring periods of quiet wakefulness, such as the sleep onset period (57, 60). However, a recent comparison study between actigraphy, PSG and sleep diaries in BDs concluded that ‘actigraphy is a valid tool for estimating sleep length and fragmentation. . .’ (18), and the present review suggests it may have many potential uses and benefits in the future. One last issue relies on the quality of the selection of ‘healthy controls’. For most studies, the selection relies on screening these subjects to exclude those with any DSM-IV axis I disorders (18, 46–52). Only one study has included ‘subjects with good sleep’ as a control group but with no mention of screening for DSM-IV axis I diagnosis (10). Given the relatively small number of eligible studies and lack of repeated reporting of the same potential confounders, the meta-regression can only be regarded as an exploratory post hoc analysis. It is quite possible that some analyses did not reach statistical significance because of a lack of statistical power. Indeed, the lack of an association between SMDs and study sample size and the duration of actigraphy recording clearly require replication after more studies come ‘on stream’.

However, in the interim, we can recommend some methodological improvements for future studies of actigraphy in remitted patients with BDs. These should at the very least include (i) careful matching of cases and controls for age, (ii) careful definition of remission and standardization of the mood symptoms scales employed and (iii) greater awareness of the potential confounding produced by heterogeneous definitions of and recruitment strategies for controls. Such modification may help reconcile some of the different findings reported across studies. This meta-analysis of actigraphy studies in remitted BDs highlights that several sleep parameters differentiate cases from controls, with consistent evidence of longer sleep latency, longer sleep duration, poorer sleep efficiency and more frequent WASO. Sleep latency and duration had moderate effect sizes (ES) and, if these findings are replicated, may help in the identification of more reliable sleep biomarkers of both susceptibility to episode onset, recovery and relapse (12, 61), which in turn may help improve the stratification of personalized approaches to treatment planning, such as consideration of the role of approaches such as CBT for insomnia in BD cases (62). To conclude, This meta-analysis of sleep in remitted BDs highlights disturbances in several sleep parameters: longer sleep latency, longer sleep duration, poorer sleep efficiency and more frequent WASO. Future actigraphy studies should pay attention to age matching and levels of residual depressive symptoms. We also recommend that future studies should consider recruiting larger samples and recording actigraphy for at least 2 weeks to improve the options for between-study comparisons of a wider range of sleep variables. Acknowledgements We are grateful to the authors, from reported and unreported studies, who provided additional information and/or replied to our queries.

Declaration of interest Pierre Alexis GEOFFROY and Mohamed LAJNEF have no conflict of interest. Jan SCOTT has received funding to attend national and international conferences, financial compensation as an independent symposium speaker for talks on early onset in BDs and psychosocial aspects of BDs, and advisory board fees from Astra Zeneca, BMS-Otsuka, Eli Lilly, GSK, Jansen-Cilag, Lundbeck, Sanofi-Aventis and Servier. Carole BOUDEBESSE has received honoraria and financial compensation as independent symposium speakers from Otsuka.

9

Geoffroy et al. Chantal HENRY has received honoraria and financial compensation as independent symposium speakers from SanofiAventis, Lundbeck, AstraZeneca, Eli Lilly and Bristol-Myers Squibb. Marion LEBOYER has received honoraria and financial compensation as an independent symposium speaker from AstraZeneca and Servier. Frank BELLIVIER has received honoraria and financial compensation as independent symposium speakers from Sanofi-Aventis, Lundbeck, AstraZeneca, Eli Lilly, Bristol-Myers Squibb and Servier. Bruno ETAIN has received honoraria and financial compensation as independent symposium speakers from SanofiAventis, Lundbeck, AstraZeneca, Eli Lilly, Bristol-Myers Squibb and Servier.

References 1. Merikangas KR, Akiskal HS, Angst J et al. Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbidity Survey replication. Arch Gen Psychiatry 2007;64:543–552. 2. Whiteford HA, Degenhardt L, Rehm J et al. Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010. Lancet 2013;382:1575–1586. 3. Tohen M, Frank E, Bowden CL et al. The International Society for Bipolar Disorders (ISBD) Task Force report on the nomenclature of course and outcome in bipolar disorders. Bipolar Disord 2009;11:453–473. 4. Lichtenstein P, Yip BH, Bj€ork C et al. Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet 2009;373:234–239. 5. Etain B, Milhiet V, Bellivier F, Leboyer M. Genetics of circadian rhythms and mood spectrum disorders. Eur Neuropsychopharmacol 2011;21(Suppl 4):S676–S682. 6. Harvey AG. Sleep and circadian rhythms in bipolar disorder: seeking synchrony, harmony, and regulation. Am J Psychiatry 2008;165:820–829. 7. McClung CA. Circadian genes, rhythms and the biology of mood disorders. Pharmacol Ther 2007;114:222–232. 8. Geoffroy PA, Bellivier F, Scott J et al. Bipolar disorder with seasonal pattern: clinical characteristics and gender influences. Chronobiol Int 2013;30:1101–1107. 9. Boudebesse C, Lajnef M, Geoffroy PA et al. Chronotypes of bipolar patients in remission: validation of the French version of the circadian type inventory in the FACE-BD sample. Chronobiol Int 2014;31:698–704. 10. Harvey AG, Schmidt DA, Scarn
a A, Semler CN, Goodwin GM. Sleep-related functioning in euthymic patients with bipolar disorder, patients with insomnia, and subjects without sleep problems. Am J Psychiatry 2005;162:50–57. 11. Rocha PMB, Neves FS, Corr^ea H. Significant sleep disturbances in euthymic bipolar patients. Compr Psychiatry 2013;54:1003–1008. 12. Sylvia LG, Dupuy JM, Ostacher MJ et al. Sleep disturbance in euthymic bipolar patients. J Psychopharmacol 2012;26:1108–1112. 13. Kaplan KA, Gruber J, Eidelman P, Talbot LS, Harvey AG. Hypersomnia in inter-episode bipolar disorder: does it have prognostic significance? J Affect Disord 2011;132:438–444. 14. Jackson A, Cavanagh J, Scott J. A systematic review of manic and depressive prodromes. J Affect Disord 2003;74:209–217.

10

15. 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;30:519–529. 16. Kaplan KA, Talbot LS, Gruber J, Harvey AG. Evaluating sleep in bipolar disorder: comparison between actigraphy, polysomnography, and sleep diary. Bipolar Disord 2012;14:870–879. 17. Boudebesse C, Geoffroy PA, Bellivier F et al. Correlations between objective and subjective sleep and circadian markers in bipolar disorder. Chronobiol Int 2014;31:698–704. 18. Millar A, Espie CA, Scott J. The sleep of remitted bipolar outpatients: a controlled naturalistic study using actigraphy. J Affect Disord 2004;80:145–153. 19. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6:e1000097. 20. Haynes PL, McQuaid JR, Ancoli-Israel S, Martin JL. Disrupting life events and the sleep-wake cycle in depression. Psychol Med 2006;36:1363–1373. 21. Burton C, McKinstry B, Szentagotai Tatar A, SerranoBlanco A, Pagliari C, Wolters M. Activity monitoring in patients with depression: a systematic review. J Affect Disord 2013;145:21–28. 22. Hedges LV, Olkin I. Statistical methods for meta-analysis. New York, NY: Academic Press, 1985. 23. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–1558. 24. Harbord R, Higgins J. Meta–regression in Stata. Stata J 2008;8:493–519. 25. Baune BT, Caliskan S, Todder D. Effects of adjunctive antidepressant therapy with quetiapine on clinical outcome, quality of sleep and daytime motor activity in patients with treatment-resistant depression. Hum Psychopharmacol 2007;22:1–9. 26. Benoit O, Royant-Parola S, Borbely AA, Tobler I, Widl€ocher D. Circadian aspects of motor activity in depressed patients. Acta Psychiatr Belg 1985;85:582–592. 27. Berle JO, Hauge ER, Oedegaard KJ, Holsten F, Fasmer OB. Actigraphic registration of motor activity reveals a more structured behavioural pattern in schizophrenia than in major depression. BMC Res Notes 2010;3:149. 28. Coffield TG, Tryon WW. Construct validation of actigraphic sleep measures in hospitalized depressed patients. Behav Sleep Med 2004;2:24–40. 29. Faurholt-Jepsen M, Brage S, Vinberg M et al. Differences in psychomotor activity in patients suffering from unipolar and bipolar affective disorder in the remitted or mild/moderate depressive state. J Affect Disord 2012;141:457–463. 30. Godfrey HP, Knight RG. The validity of actometer and speech activity measures in the assessment of depressed patients. Br J Psychiatry 1984;145:159–163. 31. Korszun A, Young EA, Engleberg NC, Brucksch CB, Greden JF, Crofford LA. Use of actigraphy for monitoring sleep and activity levels in patients with fibromyalgia and depression. J Psychosom Res 2002;52:439–443. 32. Kuhs H, Reschke D. Psychomotor activity in unipolar and bipolar depressive patients. Psychopathology 1992;25: 109–116. 33. Lemke MR, Broderick A, Zeitelberger M, Hartmann W. Motor activity and daily variation of symptom intensity in depressed patients. Neuropsychobiology 1997;36:57–61. 34. Raoux N, Benoit O, Dantchev N et al. Circadian pattern of motor activity in major depressed patients undergoing antidepressant therapy: relationship between actigraphic measures and clinical course. Psychiatry Res 1994;52:85– 98.

Actigraphy meta-analysis in bipolar disorders 35. Robillard R, Naismith SL, Rogers NL et al. Delayed sleep phase in young people with unipolar or bipolar affective disorders. J Affect Disord 2013;145:260–263. 36. Royant-Parola S, Borbely AA, Tobler I, Benoit O, Widl€ocher D. Monitoring of long-term motor activity in depressed patients. Br J Psychiatry 1986;149:288–293. 37. Todder D, Caliskan S, Baune BT. Longitudinal changes of day-time and night-time gross motor activity in clinical responders and non-responders of major depression. World J Biol Psychiatry 2009;10:276–284. 38. Todder D, Caliskan S, Baune BT. Night locomotor activity and quality of sleep in quetiapine-treated patients with depression. J Clin Psychopharmacol 2006;26:638–642. 39. Volkers AC, Tulen JHM, van den Broek WW, Bruijn JA, Passchier J, Pepplinkhuizen L. Motor activity and autonomic cardiac functioning in major depressive disorder. J Affect Disord 2003;76:23–30. 40. Volkers AC, Tulen JHM, Van Den Broek WW, Bruijn JA, Passchier J, Pepplinkhuizen L. 24-Hour motor activity after treatment with imipramine or fluvoxamine in major depressive disorder. Eur Neuropsychopharmacol 2002;12: 273–278. 41. Janney CA, Fagiolini A, Swartz HA, Jakicic JM, Holleman RG, Richardson CR. Are adults with bipolar disorder active? Objectively measured physical activity and sedentary behavior using accelerometry. J Affect Disord 2014;152–154:498–504. 42. Ensrud KE, Blackwell TL, Ancoli-Israel S et al. Use of selective serotonin reuptake inhibitors and sleep disturbances in community-dwelling older women. J Am Geriatr Soc 2006;54:1508–1515. 43. Maglione JE, Ancoli-Israel S, Peters KW et al. Depressive symptoms and subjective and objective sleep in community-dwelling older women. J Am Geriatr Soc 2012; 60:635–643. 44. Paudel ML, Taylor BC, Diem SJ et al. Association between depressive symptoms and sleep disturbances in community-dwelling older men. J Am Geriatr Soc 2008; 56:1228–1235. 45. Mullin BC, Harvey AG, Hinshaw SP. A preliminary study of sleep in adolescents with bipolar disorder, ADHD, and non-patient controls. Bipolar Disord 2011;13:425–432. 46. Jones SH, Hare DJ, Evershed K. Actigraphic assessment of circadian activity and sleep patterns in bipolar disorder. Bipolar Disord 2005;7:176–186. 47. Salvatore P, Ghidini S, Zita G et al. Circadian activity rhythm abnormalities in ill and recovered bipolar I disorder patients. Bipolar Disord 2008;10:256–265. 48. Gershon A, Thompson WK, Eidelman P, McGlinchey EL, Kaplan KA, Harvey AG. Restless pillow, ruffled mind: sleep and affect coupling in interepisode bipolar disorder. J Abnorm Psychol 2012;121:863–873. 49. Ritter PS, Marx C, Lewtschenko N et al. The characteristics of sleep in patients with manifest bipolar disorder, subjects at high risk of developing the disease and healthy controls. J Neural Transm 2012;119:1173–1184. 50. St-Amand J, Provencher MD, Belanger L, Morin CM. Sleep disturbances in bipolar disorder during remission. J Affect Disord 2012;146:112–119. 51. Geoffroy PA, Boudebesse C, Bellivier F et al. Sleep in remitted bipolar disorder: a naturalistic case-control study using actigraphy. J Affect Disord 2014;158:1–7.

52. McKenna BS, Drummond SPA, Eyler LT. Associations between circadian activity rhythms and functional brain abnormalities among euthymic bipolar patients: a preliminary study. J Affect Disord 2014;164:101–106. 53. Morgenthaler TI, Lee-Chiong T, Alessi C et al. Practice parameters for the clinical evaluation and treatment of circadian rhythm sleep disorders. An American Academy of Sleep Medicine report. Sleep 2007;30:1445–1459. 54. Cortese S, Faraone SV, Konofal E, Lecendreux M. Sleep in children with attention-deficit/hyperactivity disorder: meta-analysis of subjective and objective studies. J Am Acad Child Adolesc Psychiatry 2009;48:894–908. 55. Soreca I, Levenson J, Lotz M, Frank E, Kupfer DJ. Sleep apnea risk and clinical correlates in patients with bipolar disorder. Bipolar Disord 2012;14:672–676. 56. Geddes JR, Miklowitz DJ. Treatment of bipolar disorder. Lancet 2013;381:1672–1682. 57. Lichstein KL, Stone KC, Donaldson J et al. Actigraphy validation with insomnia. Sleep 2006;29:232–239. 58. Valli
eres A, Morin CM. Actigraphy in the assessment of insomnia. Sleep 2003;26:902–906. 59. Kushida CA, Chang A, Gadkary C, Guilleminault C, Carrillo O, Dement WC. Comparison of actigraphic, polysomnographic, and subjective assessment of sleep parameters in sleep-disordered patients. Sleep Med 2001; 2:389–396. 60. Paquet J, Kawinska A, Carrier J. Wake detection capacity of actigraphy during sleep. Sleep 2007;30:1362–1369. 61. Geoffroy PA, Leboyer M, Scott J. [Predicting bipolar disorder: what can we learn from prospective cohort studies?]. Encephale 2013; pii: S0013-7006(13)00163–2. 62. Scott J, Leboyer M, Hickie I et al. Clinical staging in psychiatry: a cross-cutting model of diagnosis with heuristic and practical value. Br J Psychiatry 2013;202:243–245.

Supporting Information Additional Supporting Information may be found in the online version of this article:

Figure S1. Actigraphic studies’ funnel plots of actigraphic parameter comparison of patients with remitted bipolar disorder versus healthy controls for a) sleep latency b) sleep duration, c) Wake After Sleep Onset (WASO), and d) sleep efficiency. Figure S2. Actigraphic studies’ SMD for sleep duration and sleep efficiency as a function of (a) duration of recording (in days), (b) total number of cases included in studies, (c) depressive symptoms (assessed by the Hamilton Depression Rating Scale- HDRS), (d) manic symptoms (assessed by the Young Mania Rating Scale- YMRS), (e) absolute age difference (difference between mean ages in Bipolar and control cases). Table S1. Details of the actigraphy monitoring device and software packages of actigraphic studies exploring remitted bipolar disorder that are included in the meta-analysis.

11