INTRODUCTION

CHILDREN'S SLEEP DISTURBANCES HAVE NOT BEENSTUDIED AS EXTENSIVELY AS THOSE OF ADULTS.Because present knowledge is largely based on parental reports,a considerable limitation is imposed, as some parents may beunaware of their child's sleep problems.1,2 Inclusion of the childas an additional informant may increase the validity of studies.However, it is unclear how valid and reliable children's reportsare and how to deal with possible discrepancies arising betweenthe two informants. Therefore, objective measurements, such aspolysomnography (PSG) or actigraphy, constitute an importantpart of sleep research in children.

During recent years, activity-based sleep assessment hasbecome an important method in sleep research. It is more cost-effective and easier to apply than PSG, the gold standard of sleepresearch, but as it is based only on motor activity and not EEGmeasurements, the resulting data is, in this respect, only an esti-mate of sleep quantity and quality. However, previous researchhas shown that sleep-wake states can actually be identified wellfrom the motor activity levels recorded by actigraphs using thealgorithm developed by Sadeh et al.3 The minute-by-minuteagreement between actigraphically derived measures and PSGscoring has been reported to be 89.9% in children and 91.2% inadolescents.3,4 Although the scoring algorithms have not estab-lished validity for waist-worn monitors, some researchers con-sider the waist to be a superior actigraph attachment site in chil-dren because of higher level of user-comfort.6,9 The risk formechanical disturbances caused for example, by humidity orblows may also decrease when the monitors are worn at thewaist.

Placement of the activity monitor can affect the recordedmean activity as reported by van Hilten et al.5 Significant differ-ences were found between the trunk and the extremities, but notbetween dominant and non-dominant wrists; sleep assessmentwas not performed. In contrast, Sadeh et al. found differencesbetween dominant and non-dominant wrists, but the sleep vari-ables remained similar.4 Glod et al. compared sleep estimatesderived from wrist- and waist-worn monitors in six children intheir study of abused children.6 They found a good correlationbetween wrist- and waist-measured sleep efficiency, but no otheranalyses were reported. We found no other studies reporting theeffect of monitor placement on sleep estimation in children.More research of this issue was called for in a recent review.7The purpose of the present study was to assess the influence ofactigraph placement on the wrist and waist on recorded activitylevels and estimated sleep variables in children.

METHODS

The study group consisted of 20 volunteers (6 boys and 14girls, mean age 10.5 years, range 7.313.3 years) recruited frompublic primary schools by advertisement. Written informed con-sent was obtained from the parents and a verbal assent from thechild. All children were medication-free, Caucasian, and ofFinnish nationality. The Mini-MotionLogger actigraphs(Ambulatory Monitoring, Inc., Ardsley, NY) were concomitantlyworn on the non-dominant wrist and the belt for a continuous 72-hour period during an ordinary school week. The parents keptlogs on monitor removal times, bedtimes (lights off), and wakingtimes (lights on) and they were given exact instructions how theadequate bedtime is defined (not to include, for example, readingtimes).

The data were analyzed by ACT2000 and AW2 software. Theraw activity data was thoroughly inspected for accuracy ofreported bedtimes and rise times as well as monitor removal peri-ods before it was scored using the algorithm developed by Sadehet al.,3 which is based on total activity counts obtained during aone-minute epoch. The down periods were defined according toparental reports and included a total of 69702 minutes. Both indi-vidual and total agreement percentages of the two sleep-wakeestimations based on wrist and waist recordings were calculatedon a minute-by-minute basis.

Actigraph Placement and Sleep Estimation in ChildrenPaavonen et alSLEEP, Vol. 25, No. 2, 2002 235

Actigraph Placement and Sleep Estimation in Children

METHODOLOGY

E Juulia Paavonen MD, Mika Fjllberg MSc, Maija-Riikka Steenari BM, and Eeva T Aronen MD, PhD

Department of Child Psychiatry, University of Helsinki, Finland

Study Objectives: To determine whether actigraph placement affectssleep estimation in children. Design: Descriptive study.Setting: Naturalistic setting.Participants: Twenty children aged 7-12 years from primary schools. Interventions: N/AMeasurements: Motor activity was measured from the waist and non-dominant wrist with actigraphs for three consecutive days during a school

week.Results: The minute-by-minute agreement of sleep-wake states betweenthe two measurement sites was 92.5%. Wrist- and waist-recorded sleepparameters correlated well and the mean values did not differ.Conclusions: Although the placement of the actigraph slightly affectedthe measured activity parameters, its influence on 3-night mean sleepestimates in children was not statistically significant. Key words: Actigraph; child; sleep

Disclosure Statement

This study was supported by the Foundation for Pediatric Research, theBiomedicum Foundation, Helsinki University Central Hospital, the FinnishCultural Foundation, and the Yrj Jahnsson Foundation.

Submitted for publication February 2001

Accepted for publication October 2001

Address correspondence to: Eeva Aronen, Department of Child Psychiatry,University of Helsinki, Lastenlinnantie 2, 00250 Helsinki, Finland; Tel: +358 9471 80223; Fax: +358 9 471 80688; E-mail: eeva.aronen@hus.fi.

Sleep estimates (sleep duration, sleep onset latency, sleep per-centage, sleep efficiency) were then averaged across the threemeasurement days and were subjected to other statistical analy-ses. Means and 95% confidence intervals of each parameter werecalculated. Differences between the two monitor attachment siteswere compared. Simple comparisons between the two place-ments were done using paired t-tests or Wilcoxon signed-ranktests. Pearson or Spearman correlation coefficients were calculat-ed. Non-parametric tests were selected if the distributions of thetwo parameters were not normal (nocturnal activity, sleep dura-tion, and sleep latency). The statistical significance level was setat p 0.1).

The mean differences between the two placements were verysmall (Table 1). However, in some cases the waist and wristrecorded sleep estimates (especially sleep duration and sleeplatency) were in disagreement. The discrepancies were consis-tently clustered in certain individuals, while among the most chil-dren all sleep estimates corresponded well.

All wrist- and waist-recorded activity and sleep variables cor-related moderately well or well. The coefficient for diurnal activ-ity was not significant (r=0.39, p=0.087), but the coefficient fornocturnal activity was (r=0.91, p

REFERENCES

1. Owens J, Spirito A, McGuinn M, Nobile C. Sleep habits and sleep dis-turbance in elementary school children. Dev Behav Ped 2000;21:27-36.2. Paavonen EJ, Aronen ET, Piha J, Tamminen T, Moilanen I, AlmqvistF. Children's sleep problems: a complementary view. Acta Paediatrica2000;89:223-8.3. Sadeh A, McGuire JPD, Urbach D, Lavie P. Actigraphically basedautomatic bedtime sleep-wake scoring: validity and clinical applica-tions. J Ambul Monitoring 1989;2:209-16.4. Sadeh A, Sharkey KM, Carskadon MA. Activity-based sleep-wakeidentification: an empirical test of methodological issues. Sleep1994;17:201-7.5. van Hilten JJ, Middelkoop HAM, Kuiper SIR, Kramer CGS, RoosRAC. Where to record motor activity: an evaluation of commonly usedsites of placement for activity monitors. Electroencephalogr ClinNeurophysiol 1993;89:359-62. 6. Glod CA, Teicher MH, Hartman CR, Harakal T. Increased nocturnalactivity and impaired sleep maintenance in abused children. J Am AcadChild Adol Psychiatry 1997;36:1236-43.7. Sadeh A, Hauri PJ, Kripke DF, Lavie P. The role of actigraphy in theevaluation of sleep disorders. Sleep 1995;18:288-302.8. Sadeh A. Evaluating night wakings in sleep disturbed infants: amethodological study of parental reports and actigraphy. Sleep1996;19:757-62.9. Teicher MH, Glod CA, Harper D, Magnus E, Brasher C, Wren F,Pahlavan K. Locomotor activity in depressed children and adolescents:I. Circadian dysregulation. J Am Acad Child Adolesc Psychiatry1993;32:760-9.

Actigraph Placement and Sleep Estimation in ChildrenPaavonen et alSLEEP, Vol. 25, No. 2, 2002 237