Failure of Timed Bright Light Exposure to Alleviate Age-Related

Sleep Maintenance Insomnia

Andrea G. Suhner, PhD, Patricia J. Murphy, PhD, and Scott S. Campbell, PhD

OBJECTIVES: To determine whether a twice-weekly maintenance schedule of evening bright light exposure is effective
in alleviating sleep maintenance insomnia on a long-term
basis, after the establishment of a more favorable phase
relationship between the core body temperature (CBT)
rhythm and sleep.
DESIGN: Subjects underwent light treatment while living
at home. Eleven to 13 consecutive days of acute light treatment (active) were followed by a 3-month maintenance
light treatment period (active or control). Subjects completed five laboratory sessions: before and after the acute
phase and once a month during the maintenance period.
SETTING: Sleep laboratory and subjects homes.
PARTICIPANTS: Fifteen older subjects (seven women,
eight men; aged 6384) with chronic (1 year) complaints
of sleep maintenance insomnia.
INTERVENTIONS: During the acute phase, all participants were exposed to evening bright light (9:00 p.m. to
11:00 p.m.; 4,000 lux). During the maintenance phase,
light treatment was reduced to a twice-weekly schedule, in
which the active group received bright light from 9:00
p.m. to 11:00 p.m. and the control group received bright
light from 3:00 p.m. to 5:00 p.m.
MEASUREMENTS: During each laboratory session, polysomnographic sleep and CBT were measured.
RESULTS: Bright light exposure during the acute treatment resulted in an average phase delay in the temperature
rhythm (Tmin) of 94 minutes. Sleep quality was not improved. No significant differences between active and control subjects were found during the maintenance phase; in
both groups, Tmin gradually reverted to the baseline
phase position, and sleep efficiency remained unchanged.

From the Laboratory of Human Chronobiology, Department of Psychiatry,

Weill Medical College of Cornell University, White Plains, New York.
This work was supported by National Institutes of Health Grants
R01MH45067, R01AG12112, R01MH54617, R01HL64581,
R01AG15370, P20MH45762, and K02MH01099.
Results of this study were presented at the 14th Annual Meeting of the Associated Professional Sleep Societies (APSS) in Las Vegas, Nevada, June 2000.
Address correspondence to Scott S. Campbell, PhD, Laboratory of Human
Chronobiology, New York Presbyterian Hospital, 21 Bloomingdale Road,
White Plains NY 10605. E-mail: sscampb@med.cornell.edu

JAGS 50:617623, 2002

2002 by the American Geriatrics Society

CONCLUSIONS: Although a significant phase shift in

CBT was achieved during the acute treatment phase, no
improvement in sleep quality was observed. Twice-weekly
light exposure was not effective in maintaining the circadian phase shift over the subsequent 3 months. Issues of
compliance and alternate etiologies for the sleep maintenance insomnia are discussed. J Am Geriatr Soc 50:617
623, 2002.
Key words: sleep disturbance; aging; circadian rhythm;
light treatment

leep maintenance insomnia has a prevalence rate of

27% to 33% in individuals aged 65 and older, making
it one of the most common sleep complaints among older
people.14 Sleep maintenance problems are characterized
by frequent and prolonged nocturnal arousals, particularly
in the second half of the night, and early morning awakenings.5,6
It has been suggested that this type of sleep disturbance is, at least in part, due to an altered phase relationship between the body core temperature (CBT) rhythm and
the timing of sleep.68 Several studies have reported that
the daily rhythm of body temperature is advanced in older
persons.811 As a result, the minimum of the temperature
rhythm (Tmin) in older individuals roughly coincides with
the midpoint of the preferred sleep period, whereas in
young subjects Tmin typically occurs toward the end of
the night. Thus, the last half of the sleep period in these
older individuals occurs on the rising portion of the temperature curve, a time in the circadian cycle during which
sleep is difficult to maintain.1214 In one study, for example, it was found that the timing of Tmin in older subjects
was significantly negatively correlated with wakefulness
after sleep onset (WASO) (i.e., the earlier Tmin, the more
WASO). Together, age and the timing of Tmin accounted
for 57% of the variance in WASO.15
Bright light exposure, timed to delay the circadian
clock and thereby reestablish an appropriate phase relationship between the temperature rhythm and the timing
of sleep, has been used successfully to alleviate sleep maintenance problems.16 In a previous study, we showed that

0002-8614/02/$15.00

618

SUHNER ET AL.

exposure to bright light (4,000 lux) for 12 consecutive

evenings delayed Tmin by more than 3 hours and significantly increased sleep efficiency in eight older subjects suffering from chronic sleep maintenance insomnia.6 Similar
results have been reported by Lack et al.,17,18 using evening
bright light treatment in early-morning-awakening insomniacs. However, it should be noted that these studies examined effects of acute bright light treatment. Because
age-related sleep maintenance insomnia is most often a
chronic condition, it is important to examine the longterm efficacy of bright light treatment. The current study
sought to evaluate such an intervention.
METHODS
Subjects
Subjects were paid volunteers recruited through various
community organizations for senior citizens and through
advertisements in local and regional newspapers. The protocol complied with the principles of the Declaration of
Helsinki and was approved by the Weill Medical College
Institutional Review Board. Written informed consent was
obtained from each subject before participation.
The participants were men and women aged 60 and
older who had experienced sleep maintenance insomnia
for at least 1 year before enrollment. All subjects were in
good general physical health as determined by a physical
examination. In addition, a brief psychiatric screening
(Hamilton Depression Rating Scale (Ham-D), 17-item version, with exclusion of item 5, which rates middle insomnia) was completed for all subjects to ensure that they
were not depressed at the time of enrollment. Subjects
with a Ham-D score of 7 or greater were excluded from
participation. Before enrollment, the nature and severity of
sleep disturbance were subjectively assessed in an initial
interview and then monitored using a 2-week sleep log. In
this log the potential subject made daily entries reporting
bedtime and wake time, number of awakenings throughout the night, frequency of napping, and several measures
of subjective sleep quality. Subjects whose logs showed evidence of sleep maintenance insomnia were enrolled, although a final decision regarding inclusion in the study
was made based on polysomnographic and circadian measures assessed during the baseline laboratory session (see
below).

APRIL 2002VOL. 50, NO. 4

JAGS

hypopneas 10/hour; number of leg movements associated with electroencephalogram-defined arousal 5/hour)
excluded the subject from further participation.
The first 2 nights served as adaptation nights, and
subjects were allowed to leave the laboratory during daytime hours to go about their normal daily activities. On
awakening in the morning after Night 3, subjects remained
in the laboratory under controlled (50 lux, 20C ambient temperature) and relatively sedentary conditions. CBT
was recorded continuously throughout the 2 evaluation
nights (Nights 3 and 4) and the intervening day using an
indwelling, rectal thermistor (YSI Series 4400, Yellow
Springs Instruments, Yellow Springs, OH) connected to an
ambulatory recording device (Minilogger, Mini Mitter, Inc.,
Bend, OR).
Immediately after the 4 nights of baseline assessment
(PRE), the acute treatment phase of the study began. For
the next 9 to 11 days, subjects underwent daily bright
light treatment while living at home. They sat in front of a
bank of lights for a 2-hour interval between 8:00 p.m. and
11:00 p.m. This interval was chosen to delay circadian
phase without interfering with (i.e., delaying) the subjects
habitual bedtimes. Three portable light boxes (30 cm  62
cm, Apollo Light Systems, Inc., Orem, UT), positioned on
each side and on top of subjects television sets such that
the light intensity was approximately 4,000 lux at eye
level when the subject was seated 1 meter away, provided
illumination. To promote maximal ocular light exposure,
subjects were instructed to watch television during the entire 2-hour treatment period and to avoid any activities
(e.g., reading or writing) that would cause them to divert

Procedures
The experimental protocol is illustrated in Figure 1. The
study was divided into two phases. First, an acute treatment phase was undertaken with the aim of establishing a
more favorable relationship between CBT and the timing
of sleep. A maintenance treatment phase was then implemented, during which an extended 12-week treatment protocol sought to maintain this phase relationship, thereby
improving sleep on a longer-term basis.
Subjects slept in the laboratory for 4 consecutive
nights. Sleep was polygraphically recorded each night and
visually scored by trained technicians using standard criteria.19 All subjects were screened for apnea and periodic leg
movements on the first night in the laboratory. Evidence
of significant sleep pathology (i.e., number of apneas or

Figure 1. Illustration of the protocol. Subjects completed five

laboratory sessions: before and after a 9- to 11-day acute phase
(PRE, POST) and once per month during the 3-month hometreatment period (M1, M2, M3). Polysomnography was recorded on all nights in the laboratory. Core body temperature
was recorded for the 36 consecutive hours at each laboratory
session. Between the PRE and POST sessions (acute phase), all
subjects underwent daily active bright light treatment (4,000
lux; 9:00 p.m. to 11:00 p.m.). From the end of the POST session through M3 (maintenance phase), subjects were assigned
to a twice-weekly schedule of active bright light (4,000 lux;
9:00 p.m. to 11:00 p.m.) or control bright light (4,000 lux;
3:00 p.m. to 5:00 p.m.).

JAGS

APRIL 2002VOL. 50, NO. 4

BRIGHT LIGHT TREATMENT FOR SLEEP MAINTENANCE INSOMNIA

their gaze from the light. A distance marker was placed on

the floor 1 meter from the television, and subjects were
strongly encouraged to position their chairs for television
viewing on the marker. Our previous studies have indicated that this distance is not inconvenient or uncomfortable for most subjects. However, compliance with this instruction was not assessed in the current study, and it is
possible that subjects sat nearer to or further away from
the light source.
Immediately after the home treatment period, subjects
returned to the laboratory for a 4-day posttreatment evaluation (POST) identical to the baseline protocol, with the
exception that they received light treatment during the
first 2 nights in the laboratory (i.e., adaptation nights).
Thus, the entire acute treatment phase continued for a total of 11 to 13 nights.
It is important to note that, throughout the entire experiment, bedtimes and wake times were not defined by the
protocol; all bedtimes and wake times during laboratory
sessions and while at home were at the subjects discretion.
After the POST evaluation session, subjects entered the
3-month maintenance phase of the protocol during which
the light treatment was reduced to a twice-weekly schedule. Subjects were randomly assigned to an active group,
which continued bright light exposure between 9:00 p.m.
and 11:00 p.m., or a control group, in which bright light
exposure occurred between 3:00 p.m. and 5:00 p.m., a time
when no significant effect on the circadian clock was expected. Subjects were unaware that the study included an
active and a control condition; instead, they were led to
believe that both treatments were active. (After debriefing
at the end of the study, subjects in the control group were
offered the active treatment). Subjects were instructed to
use the lights twice per week with 2 or 3 intervening nights
between successive treatments. The active group was also
provided with and asked to wear dark wrap-around sunglasses (UVEX welding lenses, shade 3, visual light transmission 14%) until 11:00 a.m. each day when outdoors,
to avoid light during the phase-advance portion of the
light phase response curve.2022 Compliance with wearing
the sunglasses was not assessed.
Once each month, subjects reported to the laboratory
for 3 days during which sleep and circadian phase were reassessed. Each maintenance session (M1, M2, M3) was similar to the PRE and POST protocols described above, with
the exception that only one adaptation night was required.
Throughout the entire study (acute and maintenance),
subjective sleep parameters were assessed every morning
using a questionnaire. Four 100-mm visual analog scales
(worst  0 to best  100) were used for the self-assessment of overall sleep quality, ease of falling asleep, ease of
getting up, and quality of morning wakefulness. In addition, subjects completed a daily report to indicate whether
they had used the lights the previous day.

619

Sleep Measures
The purpose of Nights 1 and 2 at the baseline and the
posttreatment session (and the first night of each maintenance session) was to adapt the subjects to the laboratory
environment. As such, data obtained on those nights were
not included in the analyses. Rather, sleep parameters
from the last 2 nights of each laboratory session were averaged and used for all subsequent analyses.
Subjective Sleep Quality
Results from daily questionnaires completed at home were
averaged for the acute treatment phase and for each of the
three maintenance intervals.
Statistical Tests
All subjects underwent active treatment during the acute
phase, and paired t tests were used to evaluate the effect of
acute daily light treatment on circadian phase and sleep
measures from the PRE to POST sessions. The efficacy of
evening light (active) versus afternoon light (control) during the maintenance phase was evaluated using a two-way
mixed analysis of variance (ANOVA) for repeated measures, with the between-subjects factor group (two
groups: active, control) and within-subjects factor time
(four time points: POST, M1, M2, M3). Paired t tests were
employed for post hoc analyses of within-group changes
where appropriate.
Pearson product moment correlations between changes
in sleep efficiency from PRE to each of the subsequent sessions (POST, M1, M2, M3) and changes in the timing of
Tmin for the same time points were calculated to determine the extent to which changes in sleep quality were related to circadian phase shifts of CBT. Likewise, the correlations between the change scores from PRE to POST, M1,
M2, and M3 of the sleep-midpoint-to-Tmin phase angle
and sleep efficiency were evaluated.
Post hoc power analyses were performed for the main
outcome measures that showed nonsignificant results, for
the purpose of assessing the sensitivity of the statistical
tests used. Group values are expressed as mean  standard deviation. Results are reported in detail when the significance level was P  .05 (two-tailed). The statistical
package used was SPSS (SPSS, Inc., Chicago, IL).
RESULTS
Results of the acute treatment phase are based on 15 subjects (seven women, eight men; aged 6384, mean age
71.5) whose sleep disturbance was polysomnographically
confirmed and who exhibited an initial temperature minimum earlier than 4:00 a.m. Fourteen (nine active, five
controls) out of the 15 subjects who completed the acute
treatment phase also completed the subsequent 3-month
maintenance protocol.

Data Analyses

Acute Phase

Temperature
Temperature data were analyzed using a complex cosine
fit with a 12-hour harmonic. The nadir, or Tmin, defined
as the point on the fitted curve at which the minimum occurs, was used as the measure of circadian phase.

Effects on Temperature
At baseline, the Tmin for the group occurred on average at
2:59 a.m. ( 59 minutes). Daily bright light exposure during the acute treatment portion of the study resulted in a
significant phase delay in Tmin of 94 minutes (  81.1)

620

SUHNER ET AL.

(t14  4.48, P  .001). Thus, at the end of the acute

phase, the average Tmin occurred at 4:33 a.m. ( 75 minutes). Nine of the 15 subjects delayed by more than 1 hour
(range: 75310 minutes), five subjects delayed by less than
1 hour (range: 1148 minutes), and one subject showed an
advance of 6 minutes. Individual temperature data before
and after the acute treatment are shown in Figure 2.
Effects on Sleep
The phase shift in Tmin was accompanied by a significant
delay of sleep onset time by an average of 44 minutes,
from 11:09 p.m. to 11:53 p.m. (t14  2.95, P  .011)
and a nonsignificant delay of wake time by an average of
32 minutes (from 6:32 a.m. to 7:04 a.m.). However, neither the time spent in bed (TOT) nor the total sleep time
(TST) changed significantly as a result of the light treatment, and no improvement in objective sleep quality was
observed. Average sleep efficiency  standard deviation
(SE  TST/TOT) at baseline was 74.2%  10.1 (range
5787%) and remained at 74.3%  9.7% (range 55
92%) after the acute treatment phase. Although seven subjects exhibited increases in sleep efficiency between 3%
and 24% from PRE to POST, only one subject reached a
sleep efficiency of over 90% after acute treatment. In addition, there were no significant changes in sleep onset latency, WASO, or percentage of sleep stages after treatment. Detailed results are presented in Table 1.
Because no significant changes from PRE to POST
were observed for any of the sleep parameters, a post hoc
power analysis was performed for SE, the main outcome
measure for sleep. Based on our previous study of light
treatment in sleep maintenance insomnia,6 it could be expected that evening bright light would lead to an average
increase in SE of 12.5%  3.8%. The power to detect
such an improvement in the present study (given the sample size of 15 subjects) was 100%.
Relationship Between Temperature and Sleep
Sleep and temperature data, summarized for each treatment group, are illustrated in Figure 3. At baseline, the average Tmin for the entire group occurred 5 minutes after

Figure 2. Timing of the temperature minimum for each subject

at baseline (PRE) and after 11 to 13 nights of evening light exposure (POST).

APRIL 2002VOL. 50, NO. 4

JAGS

the midpoint of the sleep period. Seven subjects had a

Tmin earlier than the sleep midpoint (range 8135 minutes) and eight subjects had a Tmin later than the midpoint (range 3105 minutes). Twelve of the 15 participants had a Tmin before or within 1 hour after the
midpoint of sleep at baseline.
After the acute treatment, the average Tmin occurred
69 minutes after mid-sleep, with 12 subjects having a
Tmin later than the midpoint of sleep (22360 minutes).
Thus, the phase angle between the temperature nadir and
the sleep midpoint increased by more than 1 hour (t14 
2.69, P  .018). There was no significant correlation between the change in phase angle and the change in sleep efficiency from PRE to POST. Similarly, the change scores of
Tmin and sleep efficiency between PRE and POST were
not significantly correlated.
Maintenance Phase
Effects on Temperature
Because of incomplete data sets for four subjects, this analysis was based on seven active and three control subjects.
A two-way mixed ANOVA for repeated measures revealed
that there were no significant differences in Tmin between
active and control groups (i.e., the phase delay was maintained to the same extent in both groups across the maintenance phase). For both actives and controls, the CBT
minimum showed a trend toward reversion to the baseline
phase position. At the end of the 3 months, Tmin showed
a delay of 47 minutes for the active group and 76 minutes
for the control group relative to baseline (PRE). Average
times of Tmin for the treatment groups at each maintenance phase session are shown in Figure 3.
It should be noted that post hoc power analysis (based
on the hypothesis that active treatment would maintain
two-thirds of the phase delay achieved during the acute
phase, whereas the control group would revert to the baseline phase position) revealed that, because of our small
sample size, we would have been able to detect a statistically significant difference with a power of .72.
Effects on Sleep
Sleep parameters assessed during the three maintenance
laboratory sessions are detailed in Table 1, separated by
treatment group. Three cases (two active, one control)
were excluded from the analyses because of incomplete
data sets. Repeated measures ANOVA across the maintenance phase revealed no significant differences between active (n  7) and control (n  4) subjects on
sleep parameters. By the end of the 3-month treatment
period, active subjects had a mean sleep efficiency of
77.6%  3.9, whereas control subjects showed a sleep
efficiency of 78.6%  9.6. Neither group improved significantly over baseline (PRE). Bedtimes and wake times,
which were delayed in all subjects as a result of the
acute treatment (see above), remained delayed in the active group during the maintenance phase, whereas they
reverted to pretreatment times in the control group (see
Figure 3), but this effect was not significant. Likewise,
no significant group effects or group  time interactions were found on sleep latency, WASO, or sleep stage
percentages.

JAGS

APRIL 2002VOL. 50, NO. 4

BRIGHT LIGHT TREATMENT FOR SLEEP MAINTENANCE INSOMNIA

621

Table 1. Polygraphic Sleep Parameters Assessed During the Five Sessions in the Laboratory
PRE
n  15
Parameter
Sleep efficiency, %
Sleep onset time, minutes
Wake time
Time in bed, minutes
Total sleep time, minutes
Wake after sleep onset,
minutes
Sleep latency, minutes
Wake (after sleep onset), %
Stage 1 per TST, %
Stage 2 per TST, %
Stages 3 and 4 per TST, %
REM per TST, %

POST
n  15

M1
a: n  9; c: n  5

a
c
a
c
a
c
a
c
a
c
a
c
a
c

M3
a: n  7; c: n  5

mean  standard deviation

Group
a
c
a
c
a
c
a
c
a
c

M2
a: n  8; c: n  4

74.06  9.8

74.89  8.85

2,309  01:03

2,353  00:59

0632  00:39

0703  01:02

488.2  55.5

476.0  63.6

358.8  48.5

355.3  65.7

112.0  48.8

93.9  35.7

19.8  15.9

25.8  31.0

23.06  9.2

21.00  8.1

6.23  3.87

6.75  3.16

37.09  5.97

38.04  9.58

16.95  7.29

17.13  9.14

16.67  5.80

17.08  3.95

78.62  10.88
75.51  15.02
2,409  44
2,336  31
0702  66
0704  57
467.3  43.3
446.9  74.2
366.6  54.4
340.1  108.3

75.79  6.27
79.21  6.20
2,350  36
2,307  24
0720  53
0619  50
476.2  43.6
480.1  65.8
362.6  48.9
385.3  72.0

77.64  3.85
78.63  9.57
2,417  45
2,318  62
0722  52
0642  40
458.5 41.7
457.6  74.2
357.2  38.2
359.6  70.3

67.7  40.4
97.2  59.0
29.2  27.4
10.9  6.6
15.7  9.51
22.93  15.49
6.15  3.28
5.86  2.44
40.87  9.93
43.20  17.67
18.66  9.29
12.24  6.50
19.22  6.22
15.68  6.10

87.0  22.9
83.5  22.61
23.0  22.6
13.3  8.4
19.44  6.12
18.23  4.71
8.02  4.45
6.61  4.37
39.07  9.29
34.28  9.05
15.18 7.38
23.66  6.82
19.60  4.81
17.13  3.73

76.6  20.7
76.9  31.1
18.0  9.9
7.9  5.4
19.50  8.71
17.34  8.02
7.16  2.68
7.66  5.09
41.45  10.91
40.87  15.44
16.36  8.74
18.03  12.33
18.64  4.26
16.10  4.05

Note: After the POST evaluation session, subjects were split in two groups, active (a) and control (c).
PRE, POST, M1, M2, M3  the five laboratory sessions as explained in the Methods section; TST  total sleep time; REM  rapid eye movement.

Relationship Between Temperature and Sleep

During the 3-month maintenance period, the phase angle
between the temperature rhythm and the timing of sleep
did not change differentially as a function of treatment.
There were no significant correlations between the change
in Tmin and changes in sleep efficiency from baseline
(PRE) to any of the maintenance sessions (M1, M2, M3).
Subjective Ratings of Sleep
Four visual analog scales for the assessment of subjective
sleep quality were filled out daily during the home-treatment portions of the study. Scores were summarized for
the acute treatment phase and for each of the three approximately 1-month maintenance periods. Ten subjects
(seven actives, three controls) properly completed the
questionnaires. The repeated measures ANOVA across the
four summary data points revealed no group differences
for the items ease of falling asleep, ease of getting up,
and quality of morning wakefulness. Likewise, no significant effect was found for global sleep quality when analyzed with the repeated measures ANOVA. However, a
within-group analysis revealed a significant improvement
in self-rated global sleep quality in the active group by the
end of the maintenance treatment (t6  2.96, P  .025),
whereas no significant change from baseline was reported
in the control group (Figure 4). In addition, the active
group reported significantly less difficulty falling asleep

(t6  3.79, P  .009) and a shorter subjective sleep latency

(t6  2.86, P  .035) by the end of the treatment than at
baseline; no significant changes were observed in the control group.
DISCUSSION
In this group of older subjects, evening light exposure for
11 to 13 days resulted in an average delay in Tmin of 94
minutes. Likewise, the phase angle between the temperature nadir and the midpoint of sleep increased by almost
an hour as a result of treatment. However, these circadian
phase changes had no effect on objectively measured sleep
quality.
It is possible that the 94-minute phase delay was not
sufficient to establish an acceptable phase relationship between CBT and the timing of sleep. On average, after
acute light treatment, 35% of the sleep period continued
to occur on the rising portion of the temperature curve.
Perhaps as a result, no improvement in sleep quality was
observed. In our earlier study of effects of acute bright
light treatment on sleep maintenance insomnia, Tmin was
delayed by 193 minutes, and average sleep efficiency increased from 77.5% to 90%.6 In that study, compliance
was strongly enforced by daily contact with participants.
Otherwise, the experimental protocol was virtually identical to the acute phase of the current study. Support for the
notion that a larger phase shift may have been necessary to

622

SUHNER ET AL.

Figure 3. Sleep and temperature data obtained during the five

lab sessions (PRE, POST, M1, M2, M3), summarized for each
treatment group. The bars (dark bars: active; hatched bars: control) represent the sleep period, with a white mark indicating
the midpoint of sleep. Sleep efficiencies (total sleep time/time in
bed) are given at the end of each bar. The white circles represent the temperature minima, which are connected with a solid
line for the active group and a dashed line for the control
group.

improve sleep comes from other bright light studies that

report improved sleep only after phase delays of at least 2
hours.17,18
It is likely that we could have achieved larger phase
delays by using light exposure timed to coincide more
closely with the CBT minimum. However, such a treatment regimen would have interfered with subjects usual
bedtimes. During the acute treatment, although questionnaire responses indicated good compliance in the sense
that the lights were used on most nights, we cannot be certain that subjects actually followed instructions to maintain their gaze toward the light source for the entire treatment interval. In addition, lack of compliance appears to
have been a major factor in our inability to maintain the
delay during the extended treatment period, because questionnaires showed that only a minority of subjects used
the lights as instructed. It is likely that a more stringent effort on the part of the investigators to insure compliance
would have resulted in larger phase shifts, but one important aim of the current study was to develop and test a
treatment regimen that would be more acceptable for
long-term, in-home use. Later timing for the light exposure interval and a more draconian insistence on compliance would have militated against this goal.
The failure of evening bright light treatment to alleviate sleep maintenance insomnia, even with a significant
phase delay of Tmin, may also suggest that other-thancircadian factors contributed to the sleep disturbances observed in the present study sample. That is, the sleep maintenance insomnia observed in many older subjects, although
accompanied by a phase advance of the circadian clock may
nevertheless be fundamentally different from the so-called
Familial Advanced Sleep-Phase Syndrome recently reported
to be associated with a genetic mutation.23,24
The notion that circadian rhythm abnormalities are
not the only cause of sleep maintenance problems has been
demonstrated in previous studies showing that a light-in-

APRIL 2002VOL. 50, NO. 4

JAGS

Figure 4. Subjective sleep quality (How well did you sleep?) was
assessed daily during the home-treatment portions of the study
using a visual analog scale (VAS) ranging from 0 (worst) to 100
(best). Scores were summarized for the acute treatment phase
and for each of the three approximately 1-month maintenance
periods (triangles: active (n  7); squares: control (n  3)). By
the end of the treatment (M3), self-rated sleep quality was significantly improved compared with acute in the active group
(P  .05), whereas no change was observed in the control group.

duced phase advance in young subjects results in some impairment of night-time sleep quality, but not of the severity typically observed in older subjects who exhibit a
similar phase advance.25,26 Furthermore, a study investigating the relationship between sleep and body temperature
in middle-aged and older subjects found no significant correlation between the timing of the Tmin and sleep quality
after adjustment for the influence of age.9 This was also
supported in the current study; several potential subjects
who did not meet the inclusion criterion of a Tmin earlier
than 4:00 a.m. nonetheless reported chronic sleep maintenance problems. Other evidence of multiple etiologies in
age-related sleep maintenance insomnia include recent findings that homeostatic factors influence sleep quality as a
function of aging27 and studies that indicate that comorbid
medical conditions contribute to sleep maintenance problems.4,28,29 Circadian manipulations will have limited effectiveness if noncircadian factors (e.g., psychological distress,
pain) are involved.
If the phase advance observed in many older individuals with sleep maintenance insomnia is indeed an epiphenomenon of the sleep disturbance, how can the positive results from other successful light treatment studies be
explained? Earlier, we suggested that better compliance
might be critical to achieve the degree of phase delay necessary to influence sleep quality. The possibility that a placebo effect may account for some, or all, of the improvement also cannot be eliminated.
No valid conclusion about the efficacy of the maintenance treatment regarding sleep quality can be made, because the study design was predicated on the assumption
that acute light treatment would improve sleep, which it
did not. Interestingly, subjective sleep quality improved in

JAGS

APRIL 2002VOL. 50, NO. 4

BRIGHT LIGHT TREATMENT FOR SLEEP MAINTENANCE INSOMNIA

the active group, but not in the control group, during the
maintenance treatment. This finding may be interpreted as
a placebo effect, whereby evening light exposure was perceived to be more effective than afternoon exposure. Alternatively, a nonsignificant increase in objective sleep quality
from a statistical standpoint may nevertheless be perceived
as meaningful to an individual with poor sleep.
Despite these negative findings, we believe that studies
of long-term light treatment should continue, because light
therapy may provide a safe, nonpharmacological alternative
for the treatment of age-related and other types of sleep disturbances. Improving compliance is one obvious focus for
future studies. In this regard, recent evidence indicates that
the human circadian timing system can be phase shifted
with relatively low-intensity light,30,31 and short-duration,
intermittent light pulses32 and that avoidance of light at certain circadian times is crucial for facilitating and maintaining desired changes in phase.33 It may be possible to decrease the light intensity or shorten the light treatment
duration, thereby increasing the acceptance and comfort of
light treatment for sleep maintenance insomnia.
ACKNOWLEDGMENTS
We gratefully acknowledge the technical assistance of the
staff of the Laboratory of Human Chronobiology.
REFERENCES
1. Bixler EO, Kales A, Soldatos CR et al. Prevalence of sleep disorders in the
Los Angeles metropolitan area. Am J Psychiatry 1979;136:12571262.
2. Karacan I, Thornby JI, Anch M et al. Prevalence of sleep disturbance in a primarily urban Florida County. Soc Sci Med 1976;10:239244.
3. Gislason T, Reynisdottir H, Kristbjarnarson H et al. Sleep habits and sleep
disturbances among the elderlyan epidemiological survey. J Intern Med
1993;234:3139.
4. Foley DJ, Monjan AA, Brown SL et al. Sleep complaints among elderly persons: An epidemiologic study of three communities. Sleep 1995;18:425432.
5. Webb WB, Campbell SS. Awakenings and the return to sleep in an older population. Sleep 1980;3:4146.
6. Campbell SS, Dawson D, Anderson MW. Alleviation of sleep maintenance
insomnia with timed exposure to bright light. J Am Geriatr Soc 1993;41:
829836.
7. Campbell SS, Gillin JC, Kripke DF et al. Gender differences in the circadian
temperature rhythms of healthy elderly subjects: Relationships to sleep quality. Sleep 1989;12:529536.
8. Czeisler CA, Dumont M, Duffy JF et al. Association of sleep-wake habits in
older people with changes in output of circadian pacemaker. Lancet 1992;
340:933936.
9. Campbell SS, Murphy PJ. Relationships between sleep and body temperature
in middle-aged and older subjects. J Am Geriatr Soc 1998;46:458462.

623

10. Monk TH, Buysse DJ, Reynolds CF et al. Circadian temperature rhythms of
older people. Exp Gerontol 1995;30:455474.
11. Weitzman ED, Moline ML, Czeisler CA et al. Chronobiology of aging: Temperature, sleep-wake rhythms and entrainment. Neurobiol Aging 1982;3:
299309.
12. Gillberg M, Akerstedt T. Body temperature and sleep at different times of
day. Sleep 1982;5:378388.
13. Akerstedt T. Sleepiness as a consequence of shift work. Sleep 1988;11:1734.
14. Dawson D, Campbell SS. Timed exposure to bright light improves sleep and
alertness during simulated night shifts. Sleep 1991;14:511516.
15. Zepelin H, McDonald CS. Age differences in autonomic variables during
sleep. J Gerontol 1987;42:142146.
16. Campbell SS, Terman M, Lewy AJ et al. Light treatment for sleep disorders:
Consensus report. V. Age-related disturbances. J Biol Rhythms 1995;10:
151154.
17. Lack L, Wright H. The effect of evening bright light in delaying the circadian
rhythms and lengthening the sleep of early morning awakening insomniacs.
Sleep 1993;16:436443.
18. Lack L, Schumacher K. Evening light treatment of early morning insomnia.
Sleep Res 1993;22:225.
19. Rechtschaffen A, Kales A. A Manual of Standardized Technology, Techniques and Scoring System for Sleep Stages of Human Subjects. Washington
DC: National Institutes of Health, 1968.
20. Czeisler CA, Kronauer RE, Allan JS et al. Bright light induction of strong
(type 0) resetting of the human circadian pacemaker. Science 1989;244:
13281333.
21. Honma K, Honma S. A human phase response curve for bright light pulses.
Jpn J Psych Neurol 1988;42:167168.
22. Minors DS, Waterhouse JM, Wirz-Justice A. A human phase-response curve
to light. Neurosci Lett 1991;133:3640.
23. Jones CR, Campbell SS, Zone SE et al. Familial advanced sleep-phase syndrome: A short-period circadian rhythm variant in humans. Nat Med 1999;
5:10621065.
24. Toh KL, Jones CR, He Y et al. An hPer2 phosphorylation site mutation in familial advanced sleep-phase syndrome. Science 2001;291:10401043.
25. Campbell SS, Dawson D. Aging young sleep: A test of the phase advance hypothesis of sleep disturbance in elderly. J Sleep Res 1992;1:205210.
26. Dijk DJ, Visscher CA, Bloem GM et al. Reduction of human sleep duration
after bright light exposure in the morning. Neurosci Lett 1987;73:181186.
27. Dijk DJ, Duffy JF, Riel E et al. Aging and the circadian and homeostatic regulation of human sleep during forced desynchrony of rest, melatonin and
temperature rhythms. J Physiol 1999;516:611627.
28. Wooten V. Medical causes of insomnia. In: Kryger MH, Roth T, Dement
WC, eds. Principles and practice of sleep medicine. Philadelphia, PA: W.B.
Saunders, 1989, pp 456475.
29. Gierz M, Campbell SS, Gillin JC. Sleep disturbances in various non-affective
psychiatric disorders. Psychiatr Clin North Am 1987;10:565581.
30. Boivin DB, Duffy JF, Kronauer RE et al. Dose-response relationships for resetting of human circadian clock by light. Nature 1996;379:540542.
31. Martin SK, Eastman CI. Medium-intensity light produces circadian rhythm
adaptation to simulated night-shift work. Sleep 1998;21:154165.
32. Baehr EK, Fogg LF, Eastman CL. Intermittent bright light treatment and exercise to entrain human circadian rhythms to night work. Am J Physiol 1999;
277:15981604.
33. Eastman CI, Stewart KT, Mahoney MP et al. Dark goggles and bright light
improve circadian rhythm adaptation to night-shift work. Sleep 1994;17:
535543.