actigraphic recordings in quantification of periodic leg movements during sleep in children

8
Original article Actigraphic recordings in quantification of periodic leg movements during sleep in children Hawley E. Montgomery-Downs a , Valerie M. Crabtree a , David Gozal b, * a Division of Pediatric Sleep Medicine, Department of Pediatrics, University of Louisville, 571 South Floyd Street, Suite 439, Louisville, KY 40202, USA b Department of Pediatrics, Kosair Children’s Hospital Research Institute, 570 South Preston Street, Suite 204, Louisville, KY 40202, USA Received 21 October 2004; received in revised form 2 February 2005; accepted 4 February 2005 Abstract Background: Periodic limb movement disorder (PLMD) has recently emerged as a relatively frequent and markedly underdiagnosed condition in children that induces arousals and sleep fragmentation and leads to poor learning and behavioral problems. Because a cost- effective and widely available alternative to pediatric polysomnography is needed for diagnosis of limb movement disorders, this study sought to examine whether periodic leg movements in children could be reliably identified using recently developed actigraphy software. Methods: Bilateral actigraphs were worn around the feet by 99 children ages 4–12 years during standard clinical overnight polysomnography, which included bilateral anterior tibial electromyogram (EMG). Left and right leg movements were scored independently for comparison purposes. Results: Agreement between tibial EMG and actigraphy-derived events were initially low, with movement indices being overestimated by actigraphy. This agreement was improved when a correction factor based on the average number of movements during arousals as measured by EMG was applied. However, the correction factor itself was found to differ substantially for patients who were diagnosed with PLMD compared to other patients. Conclusions: A novel actigraphic approach currently used for detection of PLM events during sleep in adults is insufficiently accurate to permit reliable estimates in children. q 2005 Elsevier B.V. All rights reserved. Keywords: Periodic leg movements; PLM; Children; Actigraphy; Sleep 1. Introduction Periodic limb movement disorder (PLMD) has recently emerged as a relatively frequent and markedly under- diagnosed condition in children. PLMD is especially prevalent among children with neurobehavioral deficits, particularly attention-deficit hyperactivity disorder (ADHD) [1–10]. In addition, children with iron deficiency [11] and renal failure [12] appear to be at risk for PLMD. Periodic limb movements of sleep (PLMS) can induce arousals and sleep fragmentation and in turn may lead to altered neurobehavioral patterns during the day. Thus, PLMD may lead to emergence of poor learning and inattentive/disruptive behaviors similar to those seen in ADHD, and such children may be misdiagnosed due to the decreased awareness of this condition by primary care practitioners [13]. Since overnight polysomnography (PSG) in the sleep laboratory, the gold-standard diagnostic method for PLMD, is costly, labor intensive, and relatively difficult to obtain by virtue of the limited number of pediatric sleep laboratories in most countries, new accurate but simple methods for the home diagnosis of PLMD in children are clearly needed. Actigraphic devices have been found to be useful for assessing sleep/wake cycles, sleep quality and maintenance, insomnia, and locomotor activity levels in children with sleep and/or psychiatric disturbances [14–17]. Several recently developed methods have enabled ambulatory monitoring of PLMS using actigraphic devices. Such approaches have been shown to be a sensitive and specific measure of PLMS in adults [1,18,19] and have been used clinically to assess drug response among restless leg Sleep Medicine 6 (2005) 325–332 www.elsevier.com/locate/sleep 1389-9457/$ - see front matter q 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.sleep.2005.02.002 * Corresponding author. Tel.: C1 502 852 2323; fax: C1 502 852 2215. E-mail address: [email protected] (D. Gozal).

Upload: hawley-e-montgomery-downs

Post on 28-Oct-2016

219 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Actigraphic recordings in quantification of periodic leg movements during sleep in children

Original article

Actigraphic recordings in quantification of periodic leg

movements during sleep in children

Hawley E. Montgomery-Downsa, Valerie M. Crabtreea, David Gozalb,*

aDivision of Pediatric Sleep Medicine, Department of Pediatrics, University of Louisville, 571 South Floyd Street, Suite 439, Louisville, KY 40202, USAbDepartment of Pediatrics, Kosair Children’s Hospital Research Institute, 570 South Preston Street, Suite 204, Louisville, KY 40202, USA

Received 21 October 2004; received in revised form 2 February 2005; accepted 4 February 2005

Abstract

Background: Periodic limb movement disorder (PLMD) has recently emerged as a relatively frequent and markedly underdiagnosed

condition in children that induces arousals and sleep fragmentation and leads to poor learning and behavioral problems. Because a cost-

effective and widely available alternative to pediatric polysomnography is needed for diagnosis of limb movement disorders, this study

sought to examine whether periodic leg movements in children could be reliably identified using recently developed actigraphy software.

Methods: Bilateral actigraphs were worn around the feet by 99 children ages 4–12 years during standard clinical overnight polysomnography,

which included bilateral anterior tibial electromyogram (EMG). Left and right leg movements were scored independently for comparison

purposes.

Results: Agreement between tibial EMG and actigraphy-derived events were initially low, with movement indices being overestimated by

actigraphy. This agreement was improved when a correction factor based on the average number of movements during arousals as measured

by EMG was applied. However, the correction factor itself was found to differ substantially for patients who were diagnosed with PLMD

compared to other patients.

Conclusions: A novel actigraphic approach currently used for detection of PLM events during sleep in adults is insufficiently accurate to

permit reliable estimates in children.

q 2005 Elsevier B.V. All rights reserved.

Keywords: Periodic leg movements; PLM; Children; Actigraphy; Sleep

1. Introduction

Periodic limb movement disorder (PLMD) has recently

emerged as a relatively frequent and markedly under-

diagnosed condition in children. PLMD is especially

prevalent among children with neurobehavioral deficits,

particularly attention-deficit hyperactivity disorder

(ADHD) [1–10]. In addition, children with iron deficiency

[11] and renal failure [12] appear to be at risk for PLMD.

Periodic limb movements of sleep (PLMS) can induce

arousals and sleep fragmentation and in turn may lead to

altered neurobehavioral patterns during the day. Thus,

PLMD may lead to emergence of poor learning and

1389-9457/$ - see front matter q 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.sleep.2005.02.002

* Corresponding author. Tel.: C1 502 852 2323; fax: C1 502 852 2215.

E-mail address: [email protected] (D. Gozal).

inattentive/disruptive behaviors similar to those seen in

ADHD, and such children may be misdiagnosed due to the

decreased awareness of this condition by primary care

practitioners [13].

Since overnight polysomnography (PSG) in the sleep

laboratory, the gold-standard diagnostic method for PLMD,

is costly, labor intensive, and relatively difficult to obtain by

virtue of the limited number of pediatric sleep laboratories

in most countries, new accurate but simple methods for the

home diagnosis of PLMD in children are clearly needed.

Actigraphic devices have been found to be useful for

assessing sleep/wake cycles, sleep quality and maintenance,

insomnia, and locomotor activity levels in children with

sleep and/or psychiatric disturbances [14–17].

Several recently developed methods have enabled

ambulatory monitoring of PLMS using actigraphic devices.

Such approaches have been shown to be a sensitive and

specific measure of PLMS in adults [1,18,19] and have been

used clinically to assess drug response among restless leg

Sleep Medicine 6 (2005) 325–332

www.elsevier.com/locate/sleep

Page 2: Actigraphic recordings in quantification of periodic leg movements during sleep in children

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332326

syndrome (RLS) patients [20,21] and population screening

[22]. However, the validity of any actigraphy device in the

diagnosis of PLMS has yet to be investigated in children.

Therefore, the objective of this prospective, cross-sectional

study was to test the validity of Actiwatchw actigraphs as an

assessment tool in measuring PLMS in children. In order to

establish its effectiveness, the unbiased readings derived

from the Actiwatchw were compared with those of

concurrent anterior tibial EMG recordings obtained during

overnight PSG in the sleep laboratory.

2. Methods

The study was approved by the Institutional Review

Boards at the University of Louisville and Kosair Children’s

Hospital; informed consent and, when appropriate, assent

for minors, were obtained.

2.1. Participants

Subjects were clinical patients ages 4–12 years referred for

overnight polysomnographic evaluation at an urban-setting

children’s hospital. Standard overnight multichannel PSG was

performed in the Sleep Medicine Center at Kosair Children’s

Hospital in a quiet, darkened room with an ambient

temperature approximately 24 8C and with a parent or

guardian present. For analyses, recording time was considered

to begin at sleep onset (09:29 p.m.G2.8 h) and end at lights on

(05:42 a.m.G0.8 h); average (SD) recording length was 8.2

(G0.13) h. No medications were used to induce sleep.

2.2. Polysomnography

The following parameters were measured: chest and

abdominal wall movement by respiratory impedance or

inductance plethysmography, heart rate by ECG, and air

flow with a sidestream end-tidal capnograph which also

provided breath-by-breath assessment of end-tidal carbon

dioxide levels (PETCO2; BCI SC-300, Menomonee Falls,

WI), a nasal pressure cannula and a thermistor. Arterial

oxygen saturation (SpO2) was assessed by pulse oximetry

(Nellcor N 100; Nellcor, Inc., Hayward, CA), with

simultaneously recorded pulse waveform. Bilateral electro-

oculogram (EOG), eight channels of electroencephalogram

(EEG), chin electromyogram (EMG), and analog output

from a body position sensor (Braebon Medical Corporation,

New York, NY) were also monitored. Tracheal sound was

monitored with a microphone sensor (Sleepmate, Mid-

lothian, VA) and digital time-synchronized video images

were collected.

2.3. Classification of sleep-disordered breathing

Snoring children with an AHI!1/h TST with no

clinically significant hypoxemia, hypercapnia, or

excessive daytime sleepiness were considered to have

primary snoring (PS). Children with clinically significant

hypoxemia, hypercapnia, daytime sleepiness, and/or

AHIO5/h TST were considered to have clinically-relevant

SDB, and were referred for surgical treatment. Those

children with AHIR1 but %5/h TST were considered to

have mild SDB, and in the vast majority were not referred

for any additional evaluation or therapy. The definition

used for non-clinically significant hypoxemia was the

absence of any events associated with oxyhemoglobin

desaturations O3% and mean SaO2R95%, while that for

hypercapnia required the absence of end-tidal carbon

dioxide tension values O50 mmHg for O3% of total

recording time.

2.4. Arousal detection

Criteria for arousal have not yet been specifically

developed for children [23]; arousals were defined as

recommended by the American Sleep Disorders Association

Task Force report [24], and included respiratory-related

(occurring immediately subsequent to an apnea, hypopnea

or snore), technician-induced, and spontaneous arousals.

The respiratory arousal index was expressed as the total

number of respiratory-related arousals per hour of TST; and

the total arousal index was expressed as the number of all

types of arousal per hour of TST.

2.5. Leg movement detection and scoring

As part of the standard PSG, bilateral anterior tibial EMG

was used to measure leg movements and all measures were

digitized using a commercially available PSG system

(Medcare Diagnostics, Buffalo, NY). Briefly, a limb move-

ment was scored if the EMG burst was between 0.5 and 5 s

in duration and was at least 25% of the burst elicited during

pre-study calibration. A PLM sequence was assigned if

there were four or more leg movements that which were

separated by at least 5, but no more than 1990 s. Scoring was

automated and then visually scored following task force

guidelines for scoring PLM [25].

For validation purposes, leg movements on EMG were

compared to those recorded using two Mini Mitter

ActiwatchwK64 Monitors. The ActiwatchwK64 is 28!27!10 mm and weighs 17.5 g. These instruments were

placed on the dorsum of each foot prior to sleep onset and

worn throughout the duration of the PSG recording. Inside

the actigraph is an accelerometric sensor that interprets

motion and presents it as a mathematical calculation. This

calculation of the patient movements is subsequently

downloaded through the use of a computerized reader.

Actigraphic leg movement scoring was automated using

Mini Mitter Actiware-PLMS software and following as

closely as possible the ASDA PLM-scoring criteria [25].

However, the Actiware software records in consecutive 2 s

epochs. This method constituted a strictly computer-derived

Page 3: Actigraphic recordings in quantification of periodic leg movements during sleep in children

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332 327

identification of PLMS which was not user-validated as are

EMG-scored movements on PSG.

The actigraphic method for determining PLM events

examines each leg independently, so for comparison

purposes scoring of anterior tibial EMG was also done

independently for left and right legs. Data from the two

recording methods were compared three different ways:

(a)

Tabl

Desc

Stan

EMG

Acti

Acti

Left and right leg movements during sleep periods were

independently scored on EMG and the resulting indices

were compared with those reported from the actigraph

software analyses.

(b)

It was initially hypothesized that the Actiwatchw would

detect fewer PLMS than the EMG, consistent with

previous findings [1,18,19], requiring a correction by

lowering the cutoff of the number of PLMS required for

a diagnosis of PLMD. However, because initial results

from the current study showed large discrepancies

between the two recording methods, with PLM indices

much higher on actigraphy, it was thought that motion

artifact during arousals and wake periods which are not

scored as leg movements on EMG may have con-

tributed to the actigraph-recorded indices. Thus, a

subsequent analysis was conducted in which EMG data

were rescored with movements during arousals and

periods of wake retained.

(c)

The final analysis was an attempt to adjust for the

over-scoring of PLMs on actigraphy. One group mean

for the number of PLM-qualifying EMG movements

per arousal for left and right legs was calculated and

then multiplied by each individual’s arousal index to

create a correction factor. The resulting left and right

leg correction factors were subtracted from the raw

actigraph-recorded left and right leg counts and

revised indices calculated. Left and right leg acti-

graph indices following correction factor application

were then compared to EMG indices in which

movements during wake were retained and those

during arousals were excluded. Thus, the correction

factor constituted a subtraction method whereby the

greater the number of arousals, the greater the amount

by which the subject’s actigraphy-scored PLM index

was reduced. It is important to note that leg

movements that caused arousals were not included

in the calculation of the number of movements

occurring during arousals.

e 1

riptive statistics for PLMI from EMG with and without movements during a

Mean Med

Left Right Lef

dard EMG 4.0 4.0 1.3

with movements in arousals and wake included 10.9 11.1 7.3

graphy 6.4 7.9 4.1

graphy with correction factor 3.3 3.1 1.4

2.6. Statistical analyses

Analysis of variance was calculated to test the difference

between diagnostic subgroups’ movement-within-arousal

indices. Independent samples t-test was used to compare

overall arousal indices between groups. Data were analyzed

using SPSS version 11.5; a P!0.05 was considered

statistically significant. Bland–Altman plots with delineated

mean difference C2 SD of the Difference (SDD) were

constructed to provide a graphic representation of agree-

ment not provided by correlation coefficients [26]. Scatter-

plots were included in figures to provide visual

representation of the data.

3. Results

One hundred sixty-eight consecutive patients attending

overnight clinical PSG who met age eligibility criteria were

invited to participate in the research study. Fifty potential

subjects declined participation with the remaining 118

agreeing to participate. Nineteen subjects were lost due to

intolerance to wearing the actigraph (7), technician error

(2), or because one (2) or both (8) of the actigraphs was later

found to have become separated from the subject. Those

who withdrew from the study did not differ with regard to

age (7.8 [2.2]), gender (42% male), ethnicity (68%

Caucasian, 32% African-American), or diagnosis (63%

sleep-disordered breathing (SDB), 32% primary snoring,

21% PLMD). Thus, 99 subjects were available for analyses.

These final subjects were 7.8 (G2.2) years of age

(4.2–12.6), 62% male, 70% Caucasian and 22% African-

American. Fifty-four percent were diagnosed with sleep-

disordered breathing, 38% had primary snoring or normal

results on PSG, and 15% were diagnosed with PLMD.

When periods of movement during arousals and wake

were excluded from EMG recordings, consistent with

standard scoring practices, the mean difference (SD of

differences (SDD)) between the EMG and actigraphy-

recorded PLMS for the left leg was 2.41 (7.31) with 12% of

subjects falling O1 SD from the mean difference and 34%

of subjects falling O.5 SD from the mean difference. For the

right leg, the mean difference (SDD) was 2.05 (5.74) with

20% falling O1 SDD from the mean difference and 41%

falling O.5 SDD from the mean difference (Table 1, Fig. 1).

rousals included, and from actigraphy with and without correction factor

ian SD IQR Range

t Right Left Right Left Right Left Right

1.5 7.4 5.8 5.1 5.8 51.0 28.0

7.9 10.7 9.0 9.1 11.2 75.5 43.6

3.9 6.1 5.5 8.7 6.9 31.1 23.7

1.3 5.7 5.3 5.1 5.6 31.5 25.1

Page 4: Actigraphic recordings in quantification of periodic leg movements during sleep in children

-60

-40

-20

0 0

20

40

60

-60

-40

-20

20

40

60

Dif

fere

nce

in P

LM

I (a

ctig

raph

y

-

EM

G)

Dif

fere

nce

in P

LM

I (a

ctig

raph

y

-

EM

G)

0

10

20

30

40

50

PLM

Ind

ex f

rom

Act

igra

phy

PLM Index from EMG0 10 20 30 40 50

(a)

Average PLMI by actigraphy and EMG Average PLMI by actigraphy and EMG

0 10 20 30 40 50 0 10 20 30 40 50

(c) (d)

PLM

Ind

ex f

rom

Act

igra

phy

0

10

20

30

40

50

PLM Index from EMG0 10 20 30 40 50

(b)

Fig. 1. (a) Scatterplot for periodic leg movement indices recorded by actigraphy or standard EMG for left leg, (b) scatterplot for periodic leg movement indices

recorded by actigraphy or standard EMG for right leg, (c) Bland–Altman plot for agreement between periodic leg movement indices measured by actigraphy or

standard EMG for the left leg, (d) Bland–Altman plot for agreement between periodic leg movement indices measured by actigraphy or standard EMG for the

right leg.

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332328

When periods of movement during arousals and wake

were included in the EMG scoring of PLM, mean difference

(SDD) for agreement for left leg was K4.5 (9.25) with 25%

falling O1 SDD from the mean difference and 40% falling

O.5 SDD from the mean difference. Mean difference (SDD)

for the right leg was K4.62 (7.03) with 21% falling O1

SDD from the mean difference and 53% falling O.5 SDD

from the mean difference (Table 1, Fig. 2).

The average (standard error of the mean (SE); range)

number of PLM-qualifying EMG movements during

arousals in this group of subjects for left and right legs

was 0.241 (.013; 0.016–0.582) and 0.243 (.013; 0.021–

0.606) movements per arousal, respectively. This value,

multiplied by the number of arousals in each subject’s

study, constituted a correction factor to eliminate the

influence of motion artifact from the actigraphy-recorded

PLMS index. Correction factor-applied actigraph-recorded

PLMS indices were compared to EMG-recorded PLMS

index from which movements during arousals (but not

wake) were excluded. With correction factor application

to actigraphy, the mean difference (SDD) for agreement

for left leg was K.69 (6.93) with 17% falling O1 SDD

from the mean difference and 34% falling O.5 SDD from

the mean difference. Mean difference (SDD) for the right

leg was K.91 (5.51) with 20% falling O1 SDD from the

mean difference and 45% falling O.5 SDD from the mean

difference (Table 1, Fig. 3).

Subjects were divided into three groups based on

diagnosis: sleep-disordered breathing (SDB), primary snor-

ing or normal (normal), or periodic limb movement disorder

(PLMD). There was a significant difference between groups

with regard to subjects’ movement-in-arousal indices used

to create the correction factor. Subjects who had been

diagnosed with PLMD had significantly higher numbers of

leg movements during arousals than those diagnosed with

clinical SDB, mild SDB, who had primary snoring, or were

normal. Average (SE) total arousal indices for PLMD-

diagnosed and other subjects were 11.8 (.74) and 13.3 (.32),

respectively, and did not differ significantly (tZ.65,

PZ.52) (Table 2).

4. Discussion

Through recording of concurrent anterior tibialis EMG

and bilateral actigraphy in a relatively large cohort of

children with various diagnostic entities, we found that

although there was a visually evident relationship between

the two recording methodologies, there was also gross

inflation of PLM events recorded with actigraphy. Agree-

ment between the two methods was assessed using Bland–

Altman plots and application of a correction factor was used

in an attempt to improve validity of this application in the

evaluation of PLMD in pediatric populations. Despite this,

Page 5: Actigraphic recordings in quantification of periodic leg movements during sleep in children

0

10

20

30

40

50

-60

-40

-20

0

20

40

60

PLM

Ind

ex f

rom

Act

igra

phy

PLM Index from EMG

Average PLMI by actigraphy and EMG

Dif

fere

nce

in P

LM

I (a

ctig

raph

y

-

EM

G)

0 10 20 30 40 50

0

10

20

30

40

50

PLM

Ind

ex f

rom

Act

igra

phy

0 10 20 30 40 50

PLM Index from EMG

0 10 20 30 40 50

(a) (b)

(c) (d)

-60

-40

-20

0

20

40

60

Average PLMI by actigraphy and EMG

Dif

fere

nce

in P

LM

I (a

ctig

raph

y

-

EM

G) 0 10 30 40 5020

Fig. 2. (a) Scatterplot for periodic leg movement indices recorded by actigraphy or EMG when movements during Arousals and Wake were included for left

leg, (b) scatterplot for periodic leg movement indices recorded by actigraphy or EMG when movements during Arousals and Wake were included for right leg,

(c) Bland–Altman plot for agreement between periodic leg movement indices measured by actigraphy, or by EMG when movements during Arousals and

Wake were included, for the left leg, (d) Agreement Bland–Altman plot for agreement between periodic leg movement indices measured by actigraphy, or by

EMG when movements during Arousals and Wake were included, for the right leg.

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332 329

the major finding of the present study is that a currently

available actigraphic approach is insufficiently accurate to

permit use of this method for reliable estimates of PLM

events in sleeping children. Indeed, visual examination of

the agreement plots indicates a triangular distribution such

that the higher the PLM index, the greater the discrepancy

between the methods.

Several methodological issues merit comment. The

population approached for the current study was typical of

the cross-sectional distribution of patients usually evaluated

at pediatric sleep centers. Indeed, there were no major

differences between the overall diagnostic characteristics of

our cohort and those of patients being evaluated prior to and

subsequent to completion of the study. Sub-analyses in

patients with PLMD failed to improve the validity of the

actigraphic method; however, even if the correction factor

had succeeded in PLMD subjects, it is important to consider

the fact that the diagnosis of PLMD involves the evaluation

of patients for whom no information about their PLM index

is known a priori. Furthermore, any reliable diagnostic

method should be able to detect the disease condition of

interest even when such is not suspected prior to testing.

These requirements were not met when the actigraphic

device approach was used, and as mentioned use of the latter

led to substantial increases in the numbers of PLM events

when compared to the tibial EMG-derived PLM indices.

This actigraphic inflation was despite the difference in

scoring epoch-length; movements less than 2 s were not

scored by the actigraph, while movements as short as .5 s

were scored on EMG. Thus, an inflation on the part of EMG

was originally expected, as has been found by others [1,18,

19].

To determine whether this inflation was due to move-

ments during arousals and wake periods, comparison was

then made when such movements were retained during

analysis of the polysomnographically recorded EMG.

Furthermore, to adjust for this motion artifact, the average

number of leg movements during arousals was used to

create a correction factor that was applied to each

subject’s actigraphically-recorded PLM index. Application

of the correction factor improved the agreement between

actigraphy and polysomnographically-derived PLM indices,

with fewer subjects falling further from the mean difference.

However, there were significant group differences in the

movement indices used to establish the correction factor.

Indeed, subjects with PLMD had seven times more leg

Page 6: Actigraphic recordings in quantification of periodic leg movements during sleep in children

-5

5

15

25

35

45

-5

5

15

25

35

45

0

PLM Index from EMG

-60

-40

-20

0

20

40

60

-10 0 10 20 30 40 50

-60

-40

-20

0

20

40

60

-10 0 10 20 30 40 50

PLM Index from EMG

PL

M I

ndex

fro

m A

ctig

raph

yD

iffer

ence

in P

LM

I (a

ctig

raph

y

- E

MG

)

Diff

eren

ce in

PL

MI

(act

igra

phy

-

EM

G)

PL

M I

ndex

fro

m A

ctig

raph

y

Average PLMI by actigraphy and EMG Average PLMI by actigraphy and EMG

0 10 20 30 40 50 10 20 30 40 50

(a) (b)

(c)(d)

Fig. 3. (a) Scatterplot for periodic leg movement indices recorded by actigraphy with application of a correction factor or standard EMG when movements

during wake (but not arousals) were included for the left leg, (b) scatterplot for periodic leg movement indices recorded by actigraphy with application of a

correction factor or standard EMG when movements during wake (but not arousals) were included for the right leg, (c) Bland–Altman plot for agreement

between periodic leg movement indices measured by actigraphy with application of a correction factor or by standard EMG when movements during wake (but

not arousals) were included, for the left leg, (d) Agreement Bland–Altman plot for agreement between periodic leg movement indices measured by actigraphy

with application of a correction factor, or by standard EMG when movements during wake (but not arousals) were included, for the right leg.

Table 2

Leg movements in arousals index for subjects with SDB, PLMD, and who

were normal

Left leg Right leg

N (subjects) Mean (SE) N (subjects) Mean (SE)

SDB 55 2.71 (.51) 55 3.64 (.82)

PLMD 10 15.79 (5.34) 10 8.90 (2.20)

Normal 34 2.65 (.77) 34 3.13 (.90)

FZ19.1 FZ4.3

P!.001 PZ.017

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332330

movements during arousal compared to children who had

SDB or to those children who had no evidence of either

SDB or PLMD.

The initial hypothesis was that a correction factor would

improve the usefulness of actigraphy in home-based

recordings of PLM in children. However, agreement between

recording methods was still low after correction factor

application; at best, the standard deviation of PLM index for

agreement was over 5. Again, without a priori knowledge of

the specific patient’s diagnosis (i.e. PLMD or no PLMD) it

would be impossible to determine the appropriate correction

factor for use, nullifying the intrinsic value of actigraphy as a

diagnostic tool. For the same reason, this device could not be

considered appropriate for the follow-up evaluation of

treatment after PSG-diagnosed PLMD in children since the

purpose of clinical treatment is to change the group into

which the subject falls (i.e. from PLMD to normal).

It is important to note that in developing the correction

factor for the current study, periods of movement during

wake were not included. After applying the correction

factor, the resulting actigraphy PLM index would include

movements during wake, and thus were compared to EMG

indices in which movements were removed during arousals,

but retained during wake. The rationale for this approach

relied on published evidence supporting the validity of

actigraphy in determination of sleep/wake cycles [27], and

therefore such detection could be used prior to analysis of

PLM events to extract periods of waking from the

subsequent derivation of the PLMS index.

The observation of increased leg movements during

arousals in children with PLMD is intriguing. We suspect

that there might be a restless leg syndrome component that

is emerging during arousals from sleep in children with

PLMD. Indeed that while they do not have increased

numbers of arousals per hour of sleep compared to other

clinical groups, their arousals may be of longer duration,

which would account for the differences in movement

indices between these subjects and those without PLMD.

We further speculate that these subjects’ increased

Page 7: Actigraphic recordings in quantification of periodic leg movements during sleep in children

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332 331

movements during arousals may be associated with

hyperactivity or attention deficit disorder or other move-

ment disorders, although the status of this set of clinical

subjects with regard to ADD/ADHD or movement disorders

is unknown.

It is important to note that EMG and actigraphy measure

different physiological events; EMG assesses electrical

activity as an output from muscle activation and

the actigraph is an accelerometer and detects generalized

movement. This may explain some of the discrepancies

between the methods but also supports the notion that

actigraphy’s use as a tool for screening for nocturnal

evidence of RLS might be a future application of this

technology.

There were four extreme outliers on the discrepancy

between the two recording methods. These were hetero-

geneous and unremarkable: three were female; two were

African-American, one Hispanic, and one White; they

ranged from 4 to 11 years of age, and had diagnoses of OSA

(2), PLMD (1), and primary snoring (1).

Our findings are inconsistent with others’ validation

studies testing actigraphic-detection of PLMS in adults.

Kazenwadel and colleagues found that the MOVOPORT

actigraphy device was accurate screening tool for PLMS [1]

and have applied this method for use in clinical drug trials for

treatment of RLS [20,21] and actigraphy has been applied to

screen for presence of PLMS in a wide age range among a

general population in the UK.[22] Two groups have

performed validation studies using the same device tested

here and have reported positive results, although neither has

reported agreement between the methods as opposed to

correlations [18,19]. We expect that this discrepancy may be

due in part to the ages of the subjects studied.

The current validation study looked at the comparison of

EMG and actigraphy recordings of leg movements on a

macro level; further examination using a more closely

synchronized approach, and possibly development of a

method whereby the actigraphic device could recognize

arousals through an increase in baseline movements is

clearly called for at this time. Further enhancement of the

validity of this method might be achieved by allowing user-

validation of raw signal as is the standard practice with

EMG. While this study suggests that further research be

performed to explore the possibility of using actigraphy to

identify PLMs in children, at this time it does not support

the method currently used for PLM detection in adults for

use in children for the purposes of diagnosing or assessing

the treatment response of PLMD.

Acknowledgements

The authors are grateful to the subjects and their families

who participated in the study. David Davis, PSGT, Renee

Ferguson, PSGT, Pam Gerges, RPSGT, Carrie Klaus,

RPSGT, Allison Parker, PSGT, and Tonya Thornton,

PSGT provided technical assistance with subject recruit-

ment and data collection.

This work was supported in part by US Department of

Education Grant H324E011001, Centers for Disease Con-

trol and Prevention Grant E11/CCE 422081-01, and

National Institutes of Health Grant F32 HL-074591. The

study was also supported by Mini Mitter Co, Inc. through

the provision of two Actiwatchesw, the Actiwatchw reader

with cable and adaptor, and the Actiwarew—Sleep software.

References

[1] Kazenwadel J, Pollmacher T, Trenkwalder C, et al. New actigraphic

assessment method for periodic leg movements (PLM). Sleep 1995;

18:689–97.

[2] Konofal E, Lecendreux M, Bouvard M, Mouren-Simeoni M. High

levels of nocturnal activity in children with attention-deficit

hyperactivity disorder: a video analysis. Psychiatry Clin Neurosci

2001;55:97–103.

[3] Picchietti DL, Walters AS. Restless legs syndrome and periodic limb

movement disorder in children and adolescents: comorbidity with

attention-deficit hyperactivity disorder. Child Adolescent Psychiatric

Clin North Am 1996;5:729–40.

[4] Picchietti DL, England SJ, Walters AS, Willis K, Verrico T. Periodic

limb movement disorder and restless legs syndrome in children with

attention-deficit-hyperactivity disorder. J Child Neuro 1998;13:

588–94.

[5] Picchietti DL, Walters AS. Moderate to severe periodic limb

movement disorder in childhood and adolescence. Sleep 1999;22:

297–300.

[6] Picchietti DL, Underwood DJ, Farris WA, et al. Further studies on

periodic limb movement disorder and restless legs syndrome in

children with attention-deficit-hyperactivity disorder. Mov Disord

1999;14:1000–7.

[7] Thorpy MJ. International classification of sleep disorders: diagnostic

and coding manual. Rochester, MN: American Sleep Disorders

Association; 1990 [p. 69–71].

[8] Chervin RD, Hedger KM. Clinical prediction of periodic leg

movements during sleep in children. Sleep Med 2001;2:501–10.

[9] Walters S, Mandelbaum DE, Lewin DS, et al. Dopaminergic therapy

in children with restless legs/periodic limb movements in sleep and

ADHD. Pediatr Neurol 2000;22:182–6.

[10] Crabtree VM, Ivanenko A, O’Brien LM, Gozal D. Periodic limb

movement disorder of sleep in children. J Sleep Res 2003;12(1):

73–81.

[11] Simakajornboon N, Gozal D, Vlasic V, et al. Periodic limb

movements in sleep and iron status in children. Sleep 2003;26(6):

735–8.

[12] Hanly PJ, Gabor JY, Chan C, Pierratos A. Daytime sleepiness in

patients with CRF: impact of nocturnal hemodialysis. Am J Kidney

Diseases 2003;41(2):403–10.

[13] O’Brien LM, Ivanenko A, Crabtree VM, et al. Sleep disturbances in

children with attention deficit hyperactivity disorder. Pediatr Res

2003;54(2):237–43.

[14] Glod CA, Teicher MH, Hartman CR, Harakal T. Increased nocturnal

activity and impaired sleep maintenance in abused children. J Am

Acad Child Adolescent Psychiatry 1997;36:1236–43.

[15] Sadeh A, Hauri PJ, Kripke DF, Lavie P. The role of actigraphy in the

evaluation of sleep disorders. Sleep 1995;18:288–302.

[16] Teicher MH. Actigraphy and motion analysis: new tools for

psychiatry. Harvard Rev Psychiatry 1995;3:18–35.

Page 8: Actigraphic recordings in quantification of periodic leg movements during sleep in children

H.E. Montgomery-Downs et al. / Sleep Medicine 6 (2005) 325–332332

[17] Teicher MH, Glod CA, Harper D, et al. Locomotor activity in

depressed children and adolescents: I circadian dysregulation. J Am

Acad Child Adolescent Psychiatry 1993;32:760–9.

[18] Sack RL, Pires ML, Brandes RW, deJongh E. Actigraphic detection of

periodic leg movements: a validation study. Sleep Abstract Suppl

2001;24:A405.

[19] Gschliesser V, Frauscher B, Kunz K, et al. Actigraphy for PLM

detection: a validation study with polysomnography. Sleep Abstract

Suppl 2004;660:A295.

[20] Trenkwalder C, Stiasny K, Pollmacher T, Wetter T, et al. pa therapy of

uremic and idiopathic restless legs syndrome: a double-blind

crossover trial. Sleep 1995;18(8):681–8.

[21] Collado-Seidel V, Kazenwadel J, Wetter TC, et al. A controlled study

of additional sr-L-dopa in L-dopa-responsive restless legs syndrome

with late-night symptoms. Neurology 1999;52(2):285–90.

[22] Morrish E, King MA, Pilsworth SN, et al. Periodic limb movement in

a community population detected by a new actigraphy technique.

Sleep Med 2002;3(6):489–95.

[23] Bandla HPR, Gozal D. Dynamic changes in EEG spectra during

obstructive apnea in children. Pediatr Pulmonol 2000;(29):359–65.

[24] Sleep Disorders Atlas Task Force. EEG arousals: scoring and rules

and examples. In: Guilleminault C, editor. Sleep, vol. 15, 1992. p.

173–84.

[25] Atlas Task Force. Recording and scoring leg movements. Sleep 1993;

16(8):748–59.

[26] Bland JM, Altman DG. Statistical methods for assessing agreement

between two methods of clinical measurement. Lancet 1986;i:

307–10.

[27] Standards of Practice Committee. Practice parameters for the use of

actigraphy in the clinical assessment of sleep disorders. Sleep 1995;

18(4):285–7.