Intraindividual Variability of Sleep: An Ongoing Systematic Review

Features of an individual's sleep/wake pattern, such as sleep timing, duration, and quality vary from night to night. Sleep across multiple days can be characterized along two dimensions: the individual mean?, and the intraindividual variability (IIV)?. This site is an ongoing effort to better understand sleep IIV, and provide information and tools for researchers to incorporate IIV as a second dimension in studying sleep/wake patterns.

This site builds on content of the first systematic review on the correlates of sleep/wake IIV, so that evidence synthesis could grow alongside the growth of new knowledge. The initial review can be accessed via open access below.

Bei, Wiley, Trinder, & Manber (2016). Beyond the mean: A systematic review on the correlates of daily intraindividual variability of sleep/wake patterns. Sleep Medicine Reviews, 28, 104-120.

Trend On this website, you will find an ongoing systematic review on the correlates of sleep IIV, and information on methodologies in researching IIV. We try to update this site regularly. A major update using the same methodology as in the initial systematic review will occur annually in December/January. The initial review covered records up to November 2014, and identified 53 articles. The systematic updates in Jan 2016, 2017, and 2018 added an additional 4, 11, and 13 studies respectively, bringing the total number of eligible studies to 81 on this website. Those added in the most recent update are marked with * in the table. Between annual updates, we will also add new studies (marked with !) that are notified via search alerts or other miscellaneous sources.

Recently published a study that fits our search criteria? Please get in touch so we can include it here.



Summary of Studies

Show/hide columns: toggle buttons below (some columns are hidden as default). To reset hide/show, reload the page.
Sort: click on the triangles on any column. Search: enter keywords in the bottom row.
Abbreviations: abbreviations are heavily used, and are listed alphabetically underneath the table. Reading highlighted parts first can substantially aid comprehension.

Author Year Design Design Details Sample Ethnicity Age %Female IIV Days IIV Measure IIV Method Theme Correlates Covariates Findings
Author Year Design Design Details Sample Ethnicity Age %Female IIV Days IIV Measure IIV Method Theme Correlates Covariates Findings
* Chan 2017 Observational 7 d actigraphy and diary, followed by computerized tasks and self-reported sleepiness and hunger self-reports. 78 Uni 76% White, 11% Black, 9% Asian, 3% Latino 20.38 (2.40) 56.0% 7 Actigraphy (with Diary) ISD B, C B: BMI;
C: delay discounting, response inhibition
Perceived stress, physical activity, daytime nap duration, alcohol/caffeine
  • BTv ~ +BMI controlling for covariates;
  • RTv ~ +BMI, NS controlling for covariates;
  • Delay discounting moderated BMI ~ BT, BTv, TSTv, RTv; associations significant in high delay discounting rate. Response disinhibition moderated BMI ~ BTv; similar pattern.
* Phillips et al 2017 Observational Campus housing. 30 d diary; on day 15, highest and lowest sleep regularity index (12 each) completed actigraphy and DLMO. 24 Uni (full sample 61) 87% White, 10% Asian, 2% Black, 1% Hawaiian/Pacific Islander 20.23 (1.27) 47.54% 30 and 15 Actigraphy (with Diary) Sleep Regularity Index (SRI): average % of being in the same state (sleep vs wake) at any 2 time points 24 h apart" A, B, C A: light exposure;
B: DLMO;
C: GPA
Respective means of sleep variables
  • High SRI (regular) ~ +TSTm (10pm-10am), -TST (10am-10pm), earlier and more variable SOT and RT, -Nap, earlier peak of sleep propensity rhythm; +light/dark cycle amplitude, +day light, +broader range of light-exposure centroid times; later DLMO; +GPA (moderate).
  • TSTm ~ SRI NS;
* Chan et al 2017 Intervention 2 w BL, 4 w weekly CBT-I or Ctrl, 2 w POST, 2 w 3-m FU. 62 Insomnia (older) 82.26% White, 6.45% Hispanic, 3.23% Black, 3.23% Asian, 4.84% Other. 69.45 (7.71) 67.74% 14 for BL, POST, FU, 28 for treatment. Actigraphy and Diary ISD A, D A: Education; D: Effects of CBT-I Respective means of sleep variables, insomnia duration, education, sleep medications.
  • CBT-I ~ +reductions in Diary SOLv and actigraphy TSTv, but not Nap;
  • Above effects mediated by reductions in BTv, RTv, TIBm;
  • Time x group x BL sleepv interaction sig: BL sleepv ~ +treatment response;
  • BL SOLv, TSTv, SEv ~ +likelihood of taking sleep medication;
  • BL SOLv ~ - lower education.
* Bernert et al 2017 Observational 3 time points with questionnaires; actigraphy and diary at T1 (1w) and T2 (2w). 50 Uni with suicide attempt history and recent ideation 74% White, 12% Hispanic, 6% Black, 2% Asian, 4% Mixed, 2% Other 19.2 (1.4) 72.0% 7 and 14 Actigraphy (with Diary) ISD A, C, D A: Shiftwork;
C: Suicidal ideation;
D: Insomnia
T1 suicide ideation, depressive symptoms, alcohol-related problems.
  • Sleep timing v ~ shiftwork;
  • SOTv ~ +change in suicidal ideation T2 to T3;
  • Sleep v ~ T2 insomnia (moderate).
* Burgess et al 2017 Observational 10 d protocol: DLMO on 2 nights both before and after a 5 d break 40 (22 DSWPD, 18 healthy) 67.5% White, 7.5% Black, 12.5% Asian, 12.5% Other DSWPD: 28.0 (7.2); Ctrl: 30.8 (7.3); 45.0% 5 Actigraphy (with Diary) RMSSD A, B, D A: morning commitments, weekday weekend difference;
B: shift in DLMO;
D: DSWPD status
None.
  • NS weekday-weekend difference in RMSSD for all sleep variables in either group;
  • Sleep v and morning commitments: NS;
  • DSWPD ~ +RTv, +TSTv, SEv and SOLv NS;
  • SOTv ~ + shift in the DLMO, but only in DSWPD.
* Bei et al 2017 Observational 7 d actigraphy and in-lab physiological measures. 436 Com 70.1% White, 28.0% Black, 1.8% Other 54.1 (11.7) 60.3% 7 Actigraphy (with Diary) Bayesian IIV model B B: allostatic load (AL) and individual system not accounted for by AL, cortisol trajectory Respective means of sleep variables. Cortisol models: age, sex, race, education, bed partner, smoking, perceived stress, chronic major medical conditions. AL analyses: age, sex, race, smoking, perceived stress, chronic major medical conditions, AL relevant medications.
  • Cortisol: flatter diurnal slope ~ +BTv (trend), +RTv, +TSTv; - awakening cortisol ~ +RTv (trend);
  • AL: controlling for age and sex: higher AL ~ +SOLv, +WASOv, +BTv (trend); these NS after controlling for other covariates.
* Cohen et al 2017 Observational Daily sleep and behavioural observations at residential facilities. 67 low-functioning autism. 88% White, 4.5% Hispanic, 4.5% Asian, 2.9% Native American 13.29 (3.06) 20.0% Average 306 (48–534). Caretakers observation every 15 or 30 min. SRI C C: Challenging behaviour, aggression, self-injury, tantrums, property destruction None.
  • Prior days' sleep predicting outcomes was most strongly driven by TSTv and nighttime wakening variability;
  • Significant predictive relationship in 81% of individuals.
* Chung 2017 Observational Actigraphy and diary study. 236 Com NR (US) 53.61 (11.67) 56.36% 7 Actigraphy and Diary Log-transformed MSSD C C: Social support, social strain Self-rated health, dyspnea, chronic conditions, depression, age, sex, marital status, exercise, caffeine, alcoholic drinks.
  • TSTv ~ + social strain but NS with social support.
* Kanady et al 2017 Intervention RCT comparing CBT-I and psychoeducation Ctrl; both 8 weekly sessions. 66 Bipolar Disorder; 47 with and 19 without comorbid insomnia diagnosis; 20 completed CBT-I, 18 completed Ctrl. Comorbid insomnia: 66.67% White, 11.11% Black, 9.26% Asian, remaining Other; Ctrl: 52.17% White, 13.04% Black, 4.35% Asian, remaining Other. Comorbid insomnia: 36.76 (11.23); Ctrl: 30.74 (10.06) Comorbid insomnia: 63.0%; Ctrl: 65.2% 7 Diary RMSSD C, D C: Working memory, verbal learning;
D: Effects of CBT-I.
Age.
  • BL TSTv ~ -verbal learning, working memory performance, independent of insomnia diagnosis;
  • Sig Group x TSTv interaction: in CBT-I group, reductions in TSTv ~ improved verbal learning (moderate).
* Breneman et al 2017 Intervention RCT on low vs high-dose exercises of 4 m; actigraphy at BL, mid-intervention (MID), POST. 49 healthy, physically inactive older women. 83.67% White, 14.29% Black, 2.04% Other 64.53 (3.83) 100% 7 at BL, MID, POST Actigraphy (with Diary) ISD, CV B B: cardiorespiratory fitness (VO₂peak), effects of exercise intervention. Mixed-effects model.
  • Main effect of time: WASOv lower at MID and POST compared to BL, but MID vs POST NS; NS for other sleep variables;
  • Using ISD: BL VO₂peak ~ -TSTv, -WASOv; Using CV, BL VO₂peak ~ -TSTv but not WASOv.
* Caia et al 2017 Observational 7 d actigraphy and diary in March (first month of competitive season). 45 Australian rugby athletes. NR (AU) 15 elite seniors: 25.5 (3.7), 15 sub-elite seniors: 22.4 (2.4), 15 elite juniors: 18.8 (0.9) NR 7 Actigraphy (with Diary) ISD A, B A: Age; B: Elite vs sub-elite None.
    Compared to seniors, juniors ~ +SOLv, +TIBv, +TSTv, +Qulityv.
* Diem et al 2016 Observational Daily Actigraphy and Diary; other cognitive and health measures. 1245 women without dementia NR (THAI) 82.6 (3.3) 100% Minimum 3 d Actigraphy (with Diary) ISD and quartile A, C A: age;
C: Risk for mild cognitive impairment (MCI) and dementia
Minimally adjusted: age, race, clinic, education. Multivariate adjusted: aforementioned, BMI, depression, comorbidities, functional impairments, smoking, alcohol, exercise, living alone, health status, antidepressant, benzodiazepine, sleep medication.
  • TSTv and SEv ~ +risk of MCI and dementia after full adjustments;
  • WASOv and SOLv NS.
* Ogilvie et al 2016 Observational 7 d actigraphy and diary; cross-sectional health measures 2146 older NR (US) 68.6 (9.2) 53.7% 7 Actigraphy (with Diary) ISD A, B A: age, sex, race
B: BMI, waist circumference, body fat
M1: age, sex, race, field center; M2: M1 plus depressive symptoms, anti-depressants, alcohol, sleep medication, smoking, income, marital status, education; M3: M2 plus sleep apnea; M4: M3 + sleep duration
  • TSTv ~ +BMI, +waist circumference, +body fat. NS controlling for demographics and TST (i.e., M4);
  • NS interactions by race, age, or sex;
  • Highest quartile TSTv had % obesity.
Aubin et al. 2016 Observational Between group comparisons on 30 D sleep 11 blind, 11 controls NR (DK) 44.5 (14.9) 63.64% 10 Actigraphy (with Diary) Range B, C B: Blindness;
C: Chronotype
None.
  • SEv and Sleep Offset v: Blind > Control.
  • SOT: NS between group.
  • Sleep on/offset timing variability and chronotype: NS.
Kaufmann et al. 2016 Observational Ecological Momentary Assessments over 11 W 41 outpatients with Bipolar I & II 78% White, 9.8% African-American, 12.2% Other 46.9 (11.8) 53.70% 77 Diary ISD and daily "atypicality" (each subject's current day TST minus TSTm then squared). A, C A: Age, sex, race, education;
C: Daily mood
None.
  • TSTv ~ -age (NS with sex, race, education); + symptom severity; - medication adherence; + variability of daily energy, sadness, & impulsivity; mean anger, anxiety, stress, & impulsivity.
  • Daily TST atypicality ~ - same day energy, + same day sadness, - anxiety in 2 days; NS for other outcomes
Tsai et al. 2016 Observational Poor vs good sleepers (based on PSQI) 197 first-time mothers in 3rd trimester NR (TW) 31.98 (4.21) 100% 7 Actigraphy (with Diary) ISD D D: Sleep complaints None.
  • TSTv, SOLv, WASOv: Poor > Good sleepers.
Lee S et al. 2016 Intervention 7 D actigraphy at BL and 1 Y FU after a 3 M workplace intervention. 396 IT employees 73.7% White, 16.2% Asian or Pacific Islander 46.97 (8.45) 41% 7 BL, 7 FU Actigraphy Within-person variance in multilevel models. A A: Age; workplace intervention Means of sleep variables.
  • Sig interaction for Age x Intervention x Time for WASOv and Napv: + age and receiving intervention ~ + reduction in FU WASOv; - age and receiving intervention ~ + increase in Napv.
  • Age x Intervention x Time NS for TSTv.
Cespedes et al. 2016 Observational 7 D actigraphy naturalistic observation 2086 Hispanics/Latinos 100% Hispanics/Latinos 47.1 (11.5) 64.77% 7 Actigraphy (with Diary) ISD; top quartile (SD > 1.5hrs) considered high IIV C/D C/D: Self-report vs actigraphy sleep duration None.
  • High TSTv ~ weaker associations between self-reported vs actigraphy TST.
Ong et al. 2016 Intervention 3-arm RCT: MBSR, MBTI, self-monitoring CTRL; daily diary for BL (1 w), EARLY (2 w), LATE (6 w), and POST intervention. 54 Insomnia NR (US) 42.9 74.10% 7 BL, 14 EARLY, 42 LATE, and 7 POST intervention Diary Location-scale mixed model D D: Treatment-related changes in SE SEm
  • Simple effects on SEv: MBSR had +reduction at EARLY, LATE, and POST; MBTI had +reduction at LATE and POST but not EARLY.
  • Compared to CTRL: MBSR had +reduction in SEv from BL to EARLY and LATE (30%), but not to POST; NS. b/w MBTI and CTRL.
  • Compared to MBTI, MBSR had +reduction in SEv from BL to EARLY (but not to LATE or POST).
Ng et al. 2016 Observational 7 d sleep diary and self-report questionnaires 84 patients with Bipolar Disorder 100% Chinese 44.04 (10.49) 67.86% 7 Diary Multilevel modeling of square successive difference A, C A: Age;
C: Social rhythm, chronotype, DBAS, sleep hygiene, major depression occurrence, type of Bipolar Disorder, depressive symptoms, mania severity.
None
  • TSTv ~ -social rhythm regularity, +major depression past 5 years;
  • SOLv and SEv ~ -social rhythm regularity;
  • WASOv ~ +DBAS, onset of major depression at 2 year follow-up;
  • Other factors NS related to any sleepv variables: age, Bipolar I or II, mania severity, depressive symptoms.
Chontong et al. 2016 Observational 5 d actigraphy and medical records extraction for correlates 41 patients with type 1 diabetes, stable insulin regimen NR (THAI) 41.5 (14.8) 61% 5 Actigraphy with Diary ISD B, C B: HbA1c, insulin requirement;
C: depressive symptoms.
Neuropathic symptoms, OSA risk, self-reported sleep quality
  • Both SMv and TSTv ~ +HbA1c;
  • High vs Low TSTv groups: High group had higher HbA1c, and higher required insulin (NS for diabetes complication or depressive symptoms).
Whiting & Murdock 2016 Observational 7 d actigraphy before in-lab cognitive tests 81 Uni 87% White, 10% Asian 20.54 (1.22) 57.80% 3 Actigraphy with Diary ISD C C: Attentional capture Means of sleep variables
  • Variable sleep interval duration ~ - attentional capture, especially for those with short sleep duration;
  • NS findings for SOTv and WTv.
Baron et al. 2016 Intervention 1 w BL , 16 w sleep hygiene with or without exercise interventions; associations between sleep variability and outcomes assessed at BL. 17 Insomnia (older, short TST) NR (US) 61.6 (4.3) 94.10% 7 for BL, 112 for Intervention Actigraphy ISD B B: Cardiometabolic disease risk (HbA1c, BMI, fasting glucose, fasting insulin, HOMA, OGTT, presleep cortisol, CRP) Age, means of sleep variables
  • TSTv ~ +HbA1c, SOTv ~ +BMI;
  • Fasting glucose, fasting insulin, HOMA, OGTT, presleep cortisol, CRP NS with sleep variability;
  • WASOv & SEv: BL > POST, but NS for TIBv, TSTv, SOTv, WTv, SOLv, Fragmentationv.
Taylor et al. 2016 Observational 14 d self-report sleep, PSG sleep, and other health measures cross-sectionally at T1; health also measured in 5.4 y (T2) 338 Com 47.6% White, 35.8% Black, 16.5% Asian 52.12 (2.10) 100% 14 Diary ISD B B: BMI, HOMA-IR BTm, TSTm, race, menopausal status, exercise, depressive symptoms; T1 values in T2 analyses; BMI in HOMA-IR analyses.
  • T1 BTv ~ +BMI (but NS after controlling for covariates) for both T1 & T2;
  • T1 BTv ~ +HOMA-IR (with/without covariates) at T1 but not T2.
Shoji et al. 2015 Observational 14 d self-report pre-sleep arousal and sleep, Young vs Old. 50 older & 50 younger Com Young: 70% White, Old: 90% White Young: 19.88 (2.76); Old: 67.81 (6.73) Young: 72%; Old: 60% 14 Diary ISD after detrending; interclass correlations A A: Age None
  • % of within- vs b/t person variability: Young > Old (SOL, WASO, TST, SQ, Pre-sleep Arousal);
  • TSTv, WASOv, SQv: Young > Old;
  • SOLv, Pre-sleep arousal IIV: NS.
Moss et al. 2015 Observational 2 w self-report, Insomnia vs Healthy 33 Insomnia, 36 Healthy Insomnia: 79% White, Healthy: 52% White Insomina: 47 (12), Healthy 32 (13) Insomina: 76%, Healthy: 73% 14 Diary (SRM) ISD D D: Insomnia Age BTv & RTv: Insomnia > Healthy
Kim et al. 2015 Observational 9 d Actigraphy and body composition via X-ray absorptiometry 191 Com (Old) 100% Japanese 83.4 (2.6) 100% 9 Actigraphy ISD for sleep variables; principle component analysis for "Inconsistent sleep/wake patterns" B B: Body composition TSTm, Napm, age, education, living alone, alcohol, smoking, Nr chorionic diseases, sleep medication, cognitive function, depression, self-reported health, sleep nights, physical activity.
  • Inconsistent sleep ~ +BMI, +%fat, +FMI, -%lean, -lean/fat ratio;
  • BTv ~ +BMI, +%fat, +FMI, -%lean, -lean/fat ratio; RTv & SMv: NS;
  • TSTv ~ +%fat, +FMI, -%lean, -lean/fat ratio; BMI: NS.
Straus et al. 2015 Observational Questionnaire, 1 w Actigraphy & Diary, PTSD vs Insomnia vs Healthy 45 PTSD, 25 Insomnia, 27 Healthy 70.1% White 33.51 (8.26) 23.70% 7 Actigraphy & Diary RMSSD C, D C: PTSD
D: Insomnia
None
  • Actigraphy & Diary TSTv, WASOv, SEv, Diary SOLv: PTSD & Insomnia both > Healthy;
  • Actigraphy SEv & Diary TSTv: PTSD > Insomnia; other variables NS b/w PTSD & Insomnia.
Ankers & Jones 2009 Observational Hypomanic risk vs Ctrl. 55 Uni NR (UK) 21.44 (3.36) 70.9% 7 Actigraphy & Diary ISD C C: Risk of hypomania (bipolar). Internal state, hypomanic interpretation
  • BTv & RTv: hypomanic risk group > Ctrl; BTv predicted group membership controlling for covariates;
  • TSTv: hypomanic risk group > Ctrl;
  • SEv & Fragmentation: hypomanic risk group > Ctrl; SOLv & %WASO: NS b/w groups.
Ari & Shulman 2012 Observational 2 time points in 1st & 2nd academic semesters. 150 Uni 88.2% Isreali 22.99 (1.75) 76.4% 7 Diary Average difference from mean. C C: Adjustment to college. None TSTv: low at both times in well-adjusted group; high in 1st low in 2nd semester in re-adjusted group.
Bijlenga et al. 2013 Observational Pts vs Ctrl 24 ADHD & DSPD vs. Ctrl NR (NL) 32.4 (10.1) 50.0% 5 Diary Variance C C: Comorbid ADHD & DSPD. None SOTv: ADHD & DSPD > Ctrl.
Bliwise et al. 2005 Observational 2 time points 10 years apart. 31 Com NR (US) 66.5 (8.0) 67.7% 14 & 7 Diary Variance A, B A: Age, cohabitation;
B: Onset of physical illness.
None
  • BTv & RTv: NS differences b/w T1 & T2 (with or without physical illness onset b/w time points); Cohabitation with -BTv & -RTv at T2.
  • Napv: NS difference b/w T1 & T2.
Bonnet & Alter 1982 Experimental 2W BL, 38 days regular sleep in lab, 4W habitual sleep, 2W FU. 12 Uni NR (US) 19-28 0.0% 14 for BL FU, 7 for manipulation. Diary ISD C, E C: Mood, vigilance, body temperature, momentary arousal. None
  • BTv, RTv, TSTv: regular < irregular conditions;
  • Mood, cognitive performance, sleep architecture: NS regular vs irregular conditions;
  • Body temperature: regular < irregular conditions.
Buman et al. 2011 Intervention RCT for sleep complaint, Exercise vs health education Ctrl. 36 Com 92% White 61.42 (6.72) NR 14 BL, 7 at 6 & 12 months after BL. Diary CV D D: Exercise for sleep complaints. None
  • BTv & RTv: NS b/w Exercise & Ctrl;
  • TIBv: Exercise < Ctrl at 6 months (moderate);
  • SOLv: Exercise < Ctrl at 12 months (moderate);
  • nWASOv: NS b/w Exercise & Ctrl.
Buysse et al. 2010 Observational Insomnia (chronic) vs Ctrl (healthy). 92 Com (older) 95.6% White 71.16 66.3% 14 Actigraphy & Diary ISD & mixed model C, D C: Depressive symptoms;
D: Insomnia symptoms.
None
  • BTv: Insomnia < Ctrl; RTv: Insomnia > Ctrl;
  • TIBv: NS Insomnia vs Ctrl; TSTv: Insomnia > Ctrl on Diary, NS on Actigraphy;
  • SEv & WASOv: Insomnia > Ctrl (Diary & Actigraphy); SOLv: Insomnia > Ctrl on Diary, NS on Actigraphy; Qualityv: Insomnia > Ctrl.
  • Insomnia group: only TIBv ~ +depressive symptoms, weak relationship b/w Diary IIV & sleep quality, sleepiness, or depressive symptoms.
Carney et al. 2006 Observational Good vs Poor sleepers based on PSQI. 243 Uni 79% White, 14% Black 20.98 (3.24) 87.9% 14 Diary (SRM) ISD D D: Good vs Poor sleep. Depressive symptoms.
  • BTv: Poor >Good sleepers, NS after controlling for depressive symptoms;
  • RTv: Poor > Good sleepers with/without controlling for depressive symptoms.
Cheek et al. 2004 Observational Women with insomnia vs healthy Ctrl. 121 Com 80% White 46.57 (4.07) 100.0% 5 Diary & PSG initiation/termination ISD D D: Insomnia symptoms Age
  • BTv ~ -sleep quality;
  • SOLv: Insomnia > Ctrl; WASOv: NS b/w Insomnia & Ctrl; Qualityv: Insomnia > Ctrl.
Dautovich et al. 2012 Observational Naturalistic observation. 103 Com (older) 96.1% White 72.90 (6.86) 64.1% 14 Actigraphy & Diary ISD after detrending based on the whole sample B B: Self-report Nr. health conditions Age Napv (Diary not Actigraphy): +Nr health conditions.
Dillon et al. 2014 Observational Naturalistic observation. 592 Normal sleepers 70.1% White, 29.9% Black 52.3 (19.5) 50.3% 14 Diary ISD & multilevel modeling A, C A: Age;
C: Depressive symptoms.
Mean values, age, sex, race, depressive symptoms.
  • TSTv: ~ +depressive symptoms, -age, age x sex, sex x race, age x sex x race (all small);
  • SOLv: ~ +SOLm (large), -age (small), +female (small), black race (small);
  • nWASOv: +nWASOm (large), +education (small), -age (small), +female (small); WASOv: +WASOm (large), -age (small);
  • Within-person variability > b/w-person variability.
Edinger et al. 1992 Intervention 2, 4, or 6W BL, 4W relaxation therapy, 2W assessment, 4W CBT, 2W assessment, 3-month FU. 7 Insomnia (sleep maintenance) 100% White 61.9 (55-68) 57.1% 14 for Diary, 7 for SAD Diary & SAD ISD D D: Insomnia intervention None
  • TIBv & TSTv: decreased over time on Diary, NS change on SAD;
  • SEv: decreased over time on both Diary & SAD;
  • SOLv: decreased over time on dairy, NS on SAD;
  • nWASOv: NS change over time on Diary; decreased after CBT on SAD, WASOv decreased over time;
  • Qualityv: NS change over time on Diary;
  • Napv: NS change over time.
  • Multivariate analyses: intervention reduced overall variability on both Diary & SAD. Most changes occurred after CBT but not after RT.
Eidelman et al. 2010 Observational Naturalistic observation of IIV followed by interviews & questionnaires. 21 Inter-episode bipolar Pts 71.4% White 37.0 (10.65) 85.7% 7 Diary ISD C C: Bipolar age of onset, lifetime manic/depressive episodes, manic/depressive symptoms. None
  • BTv: NS for all correlates (but moderate ~ +depressive episodes);
  • SOTv: ~ +depressive symptoms (large), NS ~ other correlates;
  • TSTv: NS ~ all correlates;
  • SEv: ~ +depressive episodes (large), NS ~ other correlates;
  • WASOv: ~ +depressive episode (large), NS ~ manic episodes (but moderate), NS ~ other correlates.
Espie et al. 1989 Intervention RCT of 8W relaxation, stimulus control, paradoxical intention, imagery relief placebo, or no treatment. 2W BL, assessments at 6W, 3, 6, & 17 months FU. 70 Insomnia (sleep onset) NR (UK) 44.9 (15.3) 67.1% 14 for BL, 7 for other assessments Diary with SAD ISD D D: Insomnia symptoms None
  • TSTv: ~ -time, NS treatment effect, significant treatment x time effect;
  • SOLv: ~ -time, -treatment, significant treatment x time effect;
  • Only active treatment improved sleep. Stimulus control improved sleep patterns, relaxation improved perceived sleep quality. Most results maintained at all FU.
Fainstein et al. 1997 Intervention 3 groups with sleep disturbance: with depression, with dementia, with neither. Compared day start & end IIV on 21-day melatonin treatment. 41 Insomnia NR (AR) 74 (12) 68.3% 3 Dieary CV C, D C: Cognitive & psychiatric comorbidity;
D: Sleep intervention.
None BTv: significant decrease in dementia but not other two groups.
Fichten et al. 2005 Observational 3 groups of poor (research criteria insomnia), medium, good sleeper. 148 Com (older) NR (CA) 69 (55-87) 65.5% 7 Diary ISD D D: Good vs poor sleepers & their estimation of time. None
  • TSTv: NS group differences;
  • SOLv: poor sleepers > other groups;
  • WASOv: poor sleepers > other groups;
  • Poor sleepers more variable in wake (SOL+WASO), yet this does little to explain their biased perception of own sleep problems.
Geoffroy et al. 2014 Observational Bipolar in remission vs healthy Ctrl matched on age & sex. 55 Bipolar & Ctrl NR (FR) 53.82 (10.30) 54.5% 21 Actigraphy (with Diary) ISD C C: Bipolar in remission None
  • TIBv & TSTv: Bipolar remission > Ctrl;
  • SEv: Bipolar remission > Ctrl;
  • SOLv: NS group differences;
  • WASOv: Bipolar remission > (trend) Ctrl;
  • Fragmentationv: Bipolar remission > Ctrl;
  • TSTm, SOLm, Fragmentationv, PSQI daytime function correctly classified 89% of study participants as cases or controls.
Hauri & Wisbey 1992 Observational 1W home Actigraphy then 3 days lab Actigraphy with ≥6hrs TIB. 36 Insomnia NR (US) 45 (24-69) 63.9% 7 at home 3 in lab Actigraphy ISD A A: Home vs lab setting in insomnia None TSTv: home > lab.
Hayes et al. 2014 Observational 3 groups of aMCI, non-aMCI, & "Intact". 45 Com (older) NR (US) 86.9 (4.3) 88.9% 182 Movement based bed mats (validated against Actigraphy) Inter-quartile range C C: Cognitive function None
  • WASOv: aMCI < Intact or naMCI;
  • non-aMCI showed sleep disturbance that was intermediate to that of aMCI & intact.
Hoaki et al. 2011 Observational 1W Actigraphy & questionnaires. 56 Healthy NR (JP) 26.9 (5.9) 30.4% 7 Actigraphy ISD C C: Hyperthymic temperament (bipolar) Daytime illuminance, ACTH TSTv: NS correlation with hyperthymic temperament, but ~ +hyperthymic temperament when controlling for covariates.
Innes et al. 2013 Observational 1W BL, then 1 night sleep restriction to 4 hrs, then normal sleep. 16 Healthy NR (NZ) 24.9 (20-37) 50.0% 6 Actigraphy (with Diary) ISD C C: Microsleep after sleep restriction None SOTv of BL: ~ -number of microsleeps following sleep restriction (large).
Kang & Chen 2009 Observational 2W Diary followed by assessments. 160 Uni 100% Chinese 20.3 (1.9) 49.4% 14 Diary Nr of nights/W with >1hr shift in BT: low (<1), intermediate (1-3), high (>3). C C: Subjective sleep, fatigue, sleepiness. TSTm
  • 26.9% low, 38.8% intermediate, 34.4% high.
  • BTv: with poorer subjective sleep (large, more so in intermediate & high groups); NS ~ fatigue or sleepiness.
Khawaja et al. 2013 Observational 2W Actigraphy with questionnaires. 23 Veteran with PTSD & sleep disturbance NR (US) 52.8 (10.3) 13.0% 14 Actigraphy ISD A, C A: Age;
C: PTSD.
nWASOm nWASOv: ~ -age (large), -TSTm (large).
Knutson et al. 2007 Observational 3 days’ Actigraphy, twice ~1 year apart. 669 Com 44% Black 42.9 (3.7) 57.0% 3 Actigraphy (with Diary) Custom formula A A: Race, daily vs yearly Age, race, sex, weekend TIBv, TSTv, SEv, SOLv: daily IIV > yearly IIV; black > white.
Kramer et al. 1999 Observational 2W Diary & Actigraphy in Older vs Younger. 21 (Older) & 19 (Younger) Healthy NR (NL) 65.1 (4.4) & 20.8 (2.2) 0.0% 14 Diary ISD A A: Age None
  • BTv: Younger > Older;
  • RTv: Younger > Older.
Kubo et al. 2009 Observational 1W home Diary (no naps/sleep deprivation/caffeine/alcohol); 9 days lab (Day 1 adaptation, Day 2 BL, 4 days simulated night shift, 3 days simulated day shift). 10 Healthy NR (JP) 22.9 (3.2) 0.0% 7 Diary ISD C C: Recovery pattern from simulated shift schedules None Recovery patterns related to home BTv (moderate), RTv (large), but not to alertness & performance during the simulated night shifts.
Lemola et al. 2013 Observational 1W home Actigraphy with Diary. 441 Com 66.6% White, 33.3% Black 56.85 (11.38) 60.4% 7 Actigraphy (with Diary) CV C SWL, psychological distress. Gender, age, marital status, education, BMI, ethnicity, twin status. TSTv: black >white; M < F in black; ~ -SWL, +distress (i.e., mood & anxiety).
Manber et al. 1996 Experimental 12 days BL, 4W natural vs regular sleep manipulation within 1hr window of habitual sleep timing (all asked to sleep ≥7.5hrs, light & activity upon awakening, minimize coffee), FU at 5W post for 1W 39 Uni (sleepy & irregular) NR (US) 18.8 (0.97) 69.2% 7 Diary ISD C, D, E C: Daytime sleepiness;
D: Sleep quality.
None
  • Good compliance with manipulation for regular vs natural conditions.
  • BTv: ~ +sleepiness at BL;
  • SEm increased & SOLm decreased in regular but not natural group; NS differences in WASOm;
  • When not sleep deprived, regular group had greater & longer lasting reduction in daytime sleepiness.
McBean & Montgomery-Downs 2013 Observational From beginning of 2nd postpartum week, PVT every morning for 12W. 71 Healthy (primiparas) 90.1% White 26.3 (4.1) 100.0% 7 Actigraphy & Diary ISD B, C B: Time since giving birth;
C: PVT, daytime function.
Age SMv: NS change across W2-W12; SMv at W2 ~ +PVT lapses at W2, W5-W13; NS ~ slope of change in PVT lapses over time; SMv ~ +daytime impairments.
McCrae et al. 2006 Observational 4 groups: with/without insomnia by complaint vs no complaint. Sample 1: 310, Sample 2: 103, Com (older) Sample 1: 77.7% White, 22.0% Black; Sample 2: NR 60-96 Sample 1: 51.3%, Sample 2: NR 14 Diary ISD D D: Insomnia & sleep complaints Age, education, sex, medications, health conditions BTv & RTv: NS group differences.
McCrae et al. 2012 Observational 2W Diary. 72 Com (older) “Mostly” White 70.18 (7.09) 66.7% 14 Diary ISD C C: Cognitive function (inductive reasoning, processing speed) Age, education, complaint duration TSTv & TWTv: NS ~ inductive reasoning or processing speed.
Merklinger-Gruchala et al. 2008 Observational daily self-report of TST over menstrual cycle. 95 Com 100% Polish 29.48 (3.13) 100.0% 28.88 (3.83) Diary CV A, B A: Age, education, duration of daylight;
B: Estradiol levels etc (see results).
TSTm TSTv: ~ +estradiol; NS ~ age, birth weight, education, energy intake, physical activity, weight/height, body fat, BMI, age at menarche, length of menstrual cycle during collection, mean duration of daylight.
Meyer & Maier 2006 Observational 4W SRM, comparing bipolar risk, unipolar risk, & Ctrl. 141 Uni 100% native German speaker 18.18 (2.14) 70.7% 28 SRM ISD C C: Risk for bipolar & unipolar. None TSTv: bipolar risk > both unipolar risk & Ctrl; NS b/w Ctrl & unipolar risk; results hold excluding those with unipolar or bipolar disorder.
Mezick et al. 2009 Observational 9 days Actigraphy, norepinephrine from overnight urine on nights 2 & 4. 184 Com 57.1% White, 40.8% Black 59.5 (7.2) 47.3% 9 Actigraphy ISD A, C A: Race, sex;
C: Stress;
Sex, race, age, BMI, apnea-hypopnea index, medication use, & the relevant mean
  • TSTv: NS b/w black & white, F > M; ~ +stressful life events, NS ~ norepinephrine.
  • Fragmentationv: black > white; NS b/w sexes.
  • NA as moderator: in those with higher NA, Fragmentationv (but not TSTv) ~ +stressful life events, greater TSTv & Fragmantationv ~ +norepinephrine.
Millar et al. 2004 Observational Remitted bipolar I patients vs age gender matched Ctrl. 38 remitted bipolar & Ctrl NR (UK) 46.55 (10.77) 57.9% 5 Actigraphy & Diary ISD C C: Remitted bipolar pts vs Ctrl Daily mood ratings
  • Actigraphy: NS group difference on TSTv, SOLv, SEv, WASOv (multivariate); remitted bipolar > Ctrl: TSTv & WASOv (univariate);
  • Diary: remitted bipolar > Ctrl: TSTv, SOLv, SEv, WASOv (multivariate); remitted bipolar > Ctrl: TSTv, SOLv, & SEv (univariate);
  • Best group membership model: Actigraphy TSTv, Diary TSTm, Diary SOLm.
Minors et al. 1998 Observational Diary during "typical week", comparisons of data 10 years apart. 112 Com (older) NR (UK) 73.00 80.0% 7 Diary Variance A A: Age, sex, cohabitation sex
  • BTv: T1 > T2 (NS trend) in cohabitation, NS for living alone;
  • RTv: T1 > T2 in cohabitation, NS for living alone;
  • TIBv: T1 > T2 in cohabitation, NS for living alone;
  • Napv: ~ +cohabitation at T1 but not T2; NS change b/w two times.
  • All above variables ~ -age at T1 but not T2; NS ~ sex.
Monk et al. 1991 Observational 2W Diary. 34 (Older) & 30 (Younger), Com NR (US) 83.1 (80-91) & 25.5 (21-30) 42.2% 14 Diary ISD A A: Age None BTv & RTv: Younger > Older.
Ogawa et al. 2011 Intervention 1W BL, 2W placebo. 380, Insomnia 100% Japanese 48.5 (17.0) 63.2% 7 Diary Categorical (BL SOL fluctuation <-30, +/-30, >30min groups) & ISD A, D A: Age;
D: Insomnia.
SOLm SOLv: ~ +age, +habitual SOLm, -habitual TSTm, past benzodiazepines use; significantly greater SOL reduction in those with larger BL SOL fluctuation (<30 & >30 min) compared with smaller (±30min).
Okun et al. 2011 Observational 1 or 2 wk baseline Diary. 222 Com (older) 94.1% White 73.7 (7.1) 67.1% 7 or 14 Diary ISD B, D B: Inflammation biomarkers (IL-6, TNF-α);
D: Group (Good Sleepers, Insomnia, Bereaved, Carers).
Age, sex, BMI, SF-36, depressive symptoms, stress
  • BTv: NS b/w groups; ~ +TNF-α;
  • RTv: Good Sleepers < other groups; ~ +IL-6 in Good Sleepers.
  • TIBv: Good Sleepers < Insomnia, ~ +IL-6 in Good Sleepers;
  • TSTv: Good Sleepers < other groups.
Ong et al. 2007 Observational 1W Diary at BL, comparing Morning, Intermediate, & Evening chronotypes. 312 Insomnia NR (US) 48.86 (14.08) 59.0% 7 Diary ISD C C: Chronotype in insomnia patients TWT
  • BTv: NS group difference (univariate);
  • RTv: Evening > Morning/Intermediate chronotypes (univariate);
  • Evening chronotype more variable BTv & RTv (multivariate).
Patel et al. 2014 Observational 1W Actigraphy. 6038 Com (older) NR (US) 79.91 49.4% ≥5 for M, 3 for F Actigraphy (with Diary) ISD A, B, C A: Race;
B: health outcomes;
C: Cognitive function, antidepressant use.
TSTm, demographics, mental/physical health history, antidepressants, benzodiazepines, life style factors, cognitive function.
  • SMv ~ +minority race, +diabetes, +heart failure, +antidepressants, -cognitive function, -subjective health, -TSTm, +BMI in M (not F), +obesity in F (not M);
  • TSTv ~ +minority race, +diabetes, +coronary artery disease, +heart failure, +antidepressants, -cognitive function, -subjective health, -TSTm, +BMI, +obesity.
Roane et al. 2015 Observational Diary for 9W at the start of university. 132 Uni 62.9% White, 18.9% Hispanic 18.6 (0.4) 54.0% Average 56 Diary Mean range of a 4-day moving window A, B, C A: Sex, ethnicity;
B: Weight changes;
C: Depressive symptoms, chronotype.
Sex, ethnicity, depressive symptoms, chronotype, interaction b/w sex & sleep variables
  • BTv: ~ +weight gain, +eveningness, NS ~ sex, ethnicity, depressive symptoms;
  • RTv: M >F, ~ +weight gain, +eveningness, NS ~ ethnicity, depressive symptoms;
  • TSTv: NS sex difference; significantly predicted weight gain for F but not F.
Roumelioti et al. 2010 Observational Up to 2W Diary for pts, 1W for Ctrl. 183 pts (CKD, ESRD) & Ctrl 74.3% White 52.65 37.2% 14 & 7 Diary ISD B B: CKD & ESRD Age, sex, & race Qualityv: ESRD > CKD > Ctrl; NS ~ phosphorus level, hemoglobin, bicarbonate & diabetes.
Rowe et al. 2008 Observational 1W home Actigraphy & Diary. 133 Com (older carers of dementia & noncarers) 96.2% White 72.31 (7.04) 66.3% 7 Actigraphy & Diary CV A A: Care-giving of dementia patients Age, education, depression, total Nr of medications
  • TSTv: Carers > non-carers both Actigraphy & Diary;
  • SEv: Carers > non-carers both Actigraphy & Diary;
  • SOLv: NS group difference on either Actigraphy or Diary;
  • WASOv: NS group difference on either Actigraphy or Diary;
  • Qualityv: NS group difference.
Sánchez-Ortuño & Edinger 2012 Intervention RCT on 4 biweekly CBT-I vs Sleep Hygiene for PI or CMI, POST & 6 month FU assessments. 81 Insomnia 58% White 54.2 (13.7) 12.5% 14 Actigraphy & Diary ISD D D: Type of insomnia, subjective sleep, treatment related changes. None If not specified, findings apply to both subgroups or both Actigraphy & Diary.
Total sample during BL:
  • TSTv & SEv: CMI > PI (trend); PI (not CMI) Actigraphy (not Diary) TSTv ~ +PSQI.
  • SOLv: CMI > PI (Diary not Actigraphy); NS ~ PSQI.
  • WASOv: NS CMI vs PI; ~ +PSQI in PI (not CMI).
Change based on CBT-I sample:
  • TSTv, SEv, SOLv, WASOv: BL > POST on Diary (not Actigraphy), NS POST vs FU;
  • NS b/w BL TSTv, SEv & FU PSQI; BL Actigraphy (not Diary) SOLv in PI (not CMI), & BL Actigraphy (not Diary) WASOv in CMI (not PI) ~ +PSQI at FU.
  • POST Diary SEv, SOLv, WASOv (but not TSTv) reduction ~ +reduction in PSQI in CMI (not PI).
Sánchez-Ortuño et al. 2011 Observational 2W Diary in PI or IMD. 187 Insomnia 59.7% White, 33.9% Black 47.14 (14.53) 67.7% 14 Diary MSSD D D: Subtype of insomnia Sex
  • TSTv: IMD > PI;
  • SEv, SOLv, WASOv: NS b/w PI & IMD.
Shen et al. 2008 Experimental 2W BL, 4W experimental phase for irregular participants. Experimental group: increase regularity of BT, RT, routines with review & weekly feedback; Ctrl group: learn factors affecting performance. 62 Uni (bipolar spectrum) 71.8% White, 11.3% Black 19.70 (18-24) 71.8% 14 Diary ISD C, E C: Mood lability, bipolar manic & depressive symptoms None
  • TSTv: during BL, TSTv ~ +depressive symptoms (small), higher across-day symptom variability (small); NS ~ manic symptoms, within-day symptom variability. During Experimental phase, significant decrease for both groups; decrease Experimental > Ctrl. End of Experimental phase, Experimental (large) but not Ctrl group significantly less variable than BL.
  • Increased lifestyle regularity did not result in changes in correlates.
Signal et al. 2007 Observational 1W Actigraphy & Diary at ~24W gestation (T1), 1W before delivery (T2), 1W after delivery (T3), 6W/7W postpartum (T4). 19 Healthy (pregnant) NR (NZ) 34 (29-40) 100.0% 7 Actigraphy(with Diary) ISD B B: Gestation, postpartum stage Parity All variables changed significantly over time, T3 most variable:
  • TIBv: T2, T3 > T1, T4;
  • TSTv: T3 > T1, T4;
  • SEv, WASOv, IIV of 24hr sleep episodes: T3 > T1, T2, T4.
Suh et al. 2012 Intervention 1W Diary at first & last week of a 7-session CBT-I group program. 455 Insomnia NR (US) 48 (14) 57.6% 7 Diary Composites from MSSD of sleep variables A, C, D A: Age;
C: depressive symptoms, chronotype;
D: Insomnia, treatment response.
Age Behavioural Schedule Component Score (BCS): BTv, LOv, WTv, RTv, TIBv; Insomnia Symptom Composite Score (ICS): SOLv, WASOv, TSTv.
  • BCS: ~ -age (BTv, LOv, TIBv), +eveningness (all 5 variables), +depressive symptoms (LOv, WTv, RTv, TIBv), NS ~ ISI. Independent predictors of BCS are +eveningness, +depressive symptoms, & their interaction (BCS ~ +eveningness among those with higher but not lower depressive symptoms).
  • ICS: ~ +depressive symptoms (TSTv), NS ~ ISI or chronotype.
  • CBT-I reduced IIV in all sleep variables except BT & LO. High BL BCS group had significantly higher BL & reduction in depressive symptoms. ISI decreased, but NS b/w high vs low BL BCS or high vs low BL ICS groups.
Taub & Hawkins 1979 Observational 2W sleep chart for regular vs irregular groups. 36 Uni NR (US) 18-24 0.0% 14 Sleep chart with 30min periods Categorical "regular vs irregular" based on questionnaire, confirmed on CV C C: Personality traits None Irregular group lower on: dominance, sociability, self-acceptance, self-control, achievement via conformance, & intellectual efficiency, but higher on flexibility.
Vanderlind et al. 2014 Observational 3W Actigraphy sandwiched b/w two assessment sessions (T1 & T2). 35 Uni NR (US) 19.83 (1.25) 40.0% 21 Actigraphy (with Diary) ISD C C: Depressive symptoms, subjective sleep, genes, cognitive control None TSTv: small correlation with T1 +depressive symptoms, NS ~ T2 depressive symptoms; small correlation ~ -cognitive control; NS ~ PSQI or rs11932595 (gene).
Waters et al. 2011 Observational 4W home Actigraphy with Diary comparing schizophrenia Pts & Ctrl. 13 Schizophrenia Pts & Ctrl NR (AU) 43.56 (6.48) 30.8% 28 Actigraphy (with Diary) ISD C C: Schizophrenia psychopathology None
  • TSTv: Pts > Ctrl (trend);
  • SEv & SOLv: Pts > Ctrl (trend).
Westerberg et al. 2010 Observational 2W home Actigraphy & Diary. 20 aMCI, Ctrl NR (US) 71.8 75.0% 14 Actigraphy & Diary ISD C C: Cognitive function None TIBv, TSTv, SEv, & SOLv: NS b/w groups; ~ -logical memory (small - moderate), but NS ~ 24-hr recognition.

Notes: * Studies added during the previous annual systematic updates | ! ad hoc additions between major updates | “+” and “-” indicate statistically significant positive and negative association | “<” and “>” indicate statistically lower (earlier in sleep timing) or higher (later in sleep timing) than | “~” means "associated with" | “m” and “v” attached to a sleep variable indicate the mean and IIV of this variable respectively | ACTH = adrenocorticotropic hormone | aMCI = amnesic mild cognitive impairment | b/w = between | BL = baseline | BT = bedtime | CBT-I = cognitive behavioural therapy for insomnia | CKD = chronic kidney disease | CMI = insomnia with comorbid psychiatric disorders | Com = community sample | CRP = C-reactive protein | Ctrl = control group | CV = coefficient of variation | DSPD = delayed sleep phase disorder | Effect sizes: wherever reported, effect sizes were reported as small (.1 ≤ r <.3 or .2 ≤ Cohen’s d < .5), moderate (.3 ≤ r <.5 or .5 ≤ Cohen’s d < .8), large (r ≥ .5 or Cohen’s d ≥ .8) | ESRD = end stage renal disease | FMI = fat mass index | Fragmentation = Actigraphy fragmentation index | FU = follow-up | HOMA-IR = homeostatic model assessment-insulin resistance | IIV = intraindividual variability | IMD = insomnia related to mental disorder | ISD = individual standard deviation | ISI = Insomnia Severity Index | LO = lights out | M = male | MBSR = mindfulness-based stress reduction | MBTI = mindfulness-based therapy for insomnia | MSSD = Mean Square of Successive Difference | NA = negative affect | Nap = daily nap time | NR = not reported | Nr. = Number of | NS = non-significant | OGTT = oral glucose tolerance test | PI = primary insomnia | POST = post-intervention | PSQI = total scores of Pittsburgh Sleep Quality Index | Pts = patients | PVT = psychomotor vigilance task | Quality = sleep diary subjective sleep quality | RCT = randomized controlled trial | RMSSD = Root Mean Square of the Successive Differences | RT = rise-time | SAD = Sleep Assessment Device (Somtrak ST-100) | SD = standard deviation | SE = sleep efficiency | SF-36 = 36 item Short Form Survey | SM = Sleep Midpoint | SOL = sleep onset latency | SOT = time of sleep onset | SRI = Sleep Regularity Index | SRM = Social Rhythm Metric | SWL = satisfaction with life | T1, T2, … = the 1st, 2nd, … time point | TIB = time-in-bed | TST = total sleep time | TWT = total wake time | Uni = university students | WASO = wake after sleep onset, with n and % signifying the number or percentage | WT = time of awakening | x: by, or variable interaction.



Themes

This section briefly summarizes findings on how sleep IIV is related to different correlates, and are presented in the following 5 themes. These summaries are intended for "a quick glance" only. To locate detailed summaries on specific studies, including their methodologies, please search the "Theme" or "Correlates" column of the table.

This section does not include studies that are added ad hoc between annual updates (with ! in the table).

A. Demographic and environmental factors (n = 26)

Age (n = 16): Overall, in community samples, younger age was associated with more variable self-report sleep timing, duration, and quality. Over a 10-y period, sleep timing IIV did not change significantly among middle aged or older adults who lived alone, but reduced among older adults who cohabitated.
Sex (n = 5): Each study targeted a different age group, and findings on the association between sex and sleep/wake IIV were inconclusive.
Race/ethnicity (n = 7): These studies examined different samples. Overall, minority race was associated with more variable sleep quality, and findings on sleep timing and duration are mixed.
Cohabitation (n = 2): In mid-old aged community adults, cohabitation was associated with less variable sleep timing and duration.
Education (n = 3): 2 studies found non-significant association between sleep duration IIV and education; 1 found more variable SOL associated with lower education.
Care-giving to dementia patients was associated with significantly more variable sleep duration and quality. Home environment was associated with more variable sleep duration a sleep laboratory. Shiftwork, weekend, but not morning commitments are associated with more variable sleep timing. Light exposure has been shown to differ between individuals with higher and lower sleep IIV.


Methods for Quantifying IIV

Following considerations are necessary When choosing analytic method for sleep IIV:

  1. Does it account for systematic trends? Methods that do not account for systemic trends (e.g., therapeutic effects of a sleep intervention, effects of seasonal variation in daylight on sleep timing) might inflate the estimation of IIV if such trends are not of direct relevance to the hypothesis.
  2. Does it address measurement error? What's the required sample size and number of repeated measures to achieve reliable measure of IIV?
  3. Does it address missing data? Missing data is common in daily measures. Analytic approaches that address missing data are more appropriate than listwise deletion or omission.
  4. What are the covariates to include? We recommend that the individual mean of the sleep variable be included as a covariate when the effects of the IIV are examined. This helps assess the distinctive effects of IIV beyond the mean values. This is particularly recommended for sleep variables with skewed distribution (e.g., SOL), such that individuals with more extreme means have smaller IIV due to floor and ceiling effects. Compared to variables that are typically normally distributed (e.g., sleep duration and timing), skewed variables are more likely to be affected by this effect as correlations between the individual mean and IIV are likely to be higher in these variables.

Please click on the following tabs on information regarding specific IIV methodologies.


Formula Based Calculations

The individual standard deviation (ISD) and coefficient of variation (CV) are the most commonly used methods by existing studies. Others include, variance, range, interquartile range, mean successive squared difference (MSSD), and root mean squared deviation (RMSD).

Limitations:
  1. All above mentioned approaches do not account for measurement error and are susceptible to low reliability, particularly when the number of repeated observations and/or individual differences in IIV is small?. Unlike means, which have good reliability with a few repeated measures, if IIV were to be estimated using ISD, as many as 50 repeated observations would be required for a reliable instrument (reliability .9) to achieve reasonable reliability (reliability .8)?.
  2. These methods do not address missing data. This can lead to reduction in usable sample sizes (in cases of listwise deletion) and biases towards days with no missing data (in cases of omission).
  3. ISD, variance, range, and CV, may inflate the estimation of IIV by including systematic time effects. Methods such as MSSD are less sensitive to such effects, but have significant limitations when missing data is present.

About/Contact Us

The initial systematic review was conducted by Bei Bei and Joshua F Wiley from Monash University (AU), John Trinder from University of Melbourne (AU), and Rachel Manber from Stanford University (US). This website is maintained by Bei Bei. The authors would like to thank Lin Shen (2017 and 2018 update) and Zhi Xiang On (2016 update) for updating the systematic search.

To cite this website, please cite the initial publication. Questions, comments, or tell us about your study, please email bei.bei@monash.edu.


Last Updated: April 2018 | Copyright © 2018 The Authors