Sleep Biol.

Rhythms (2017) 15:159–166

DOI 10.1007/s41105-017-0095-9

ORIGINAL ARTICLE

The effects of kiwi fruit consumption in students with chronic

insomnia symptoms: a randomized controlled trial
Øystein Ottesen Nødtvedt1   · Anita Lill Hansen2,3 · Bjørn Bjorvatn4,5 ·
Ståle Pallesen1,4 

Received: 28 October 2016 / Accepted: 22 February 2017 / Published online: 6 March 2017
© Japanese Society of Sleep Research 2017

Abstract  Insomnia is the most common sleep disorder. 12 outcome variables (sleep quality and daytime function-
Although treatments such as cognitive-behavioral therapy ing as reported using sleep diary), there was a statistically
have been shown to be effective, there are limitations in significant group  ×  time interaction effect favoring the
terms of effects, accessibility, and cost. It is thus of inter- kiwi condition compared to pear. Although there were no
est to supplement treatment with more accessible means to such effects using objective measures, the results suggest
increase treatment effects. Little research exists concerning that kiwi may possess some sleep improving properties.
the effects of nutrition on sleep. Kiwi fruit contains rich Strengths and limitations of the study are discussed.
levels of nutrients, such as antioxidants, flavonoids, carot-
enoids, anthocyanins, folate, and melatonin, all of which Keywords  Insomnia · Nutrition · Sleep · Kiwi · Fruit ·
could possibly facilitate sleep. Thus, the purpose of this Randomized controlled trial
study was to investigate whether kiwi had beneficial effects
on sleep compared to a control fruit chosen on the basis
of differences in relevant nutritional content. In this rand- Introduction
omized controlled trial, 74 students suffering from chronic
insomnia symptoms were instructed to ingest either 130 g Insomnia is commonly defined as subjective reports of
of kiwi or pear, the latter comprising the control condition, poor sleep with accompanying impairments in daytime
1 h before bedtime every day for 4 weeks following 1 week functioning [1]. According to the criteria found in the 5th
of baseline assessment. Outcome measures consisted of edition of the Diagnostic and Statistical Manual of Mental
sleep diaries and actigraphy. In addition, we administered Disorders, insomnia comprises insufficient sleep, inter-
the Bergen Insomnia Scale and Pittsburgh Sleep Question- rupted sleep, and difficulties in initiating or maintaining
naire Index. Results showed that on a total of two out of sleep [2]. When defined according to formal diagnostic cri-
teria, chronic insomnia seems to have a prevalence of about
6–15% in the general population [3–5]. In addition to being
* Øystein Ottesen Nødtvedt the most common sleep disorder, a Norwegian study sug-
[email protected]
gests an increase in the prevalence during the last decade
1
Department of Psychosocial Science, University of Bergen, [5]. Longitudinal studies have shown that insomnia, even
Christiesgt. 12, 5015 Bergen, Norway after adjusting for a variety of confounders, is a predictor,
2
Department of Clinical Psychology, University of Bergen, among other things, of sick leave [6], disability pension [7],
Bergen, Norway depression [8], and even mortality [9]. Although interven-
3
Centre for Research and Education in Forensic Psychiatry, tions such as cognitive-behavioral therapy have been shown
Haukeland University Hospital, Bergen, Norway to be effective both the short and long terms [10, 11], such
4
Norwegian Competence Center for Sleep Disorders, therapy is not readily available and it is often relatively
Haukeland University Hospital, Bergen, Norway demanding in terms of therapy resources [12]. In addition,
5
Department of Global Public Health and Primary Care, compared to cognitive-behavioral therapy for other disor-
University of Bergen, Bergen, Norway ders, the treatment effects related to insomnia are relatively

13
Vol.:(0123456789)

160 Sleep Biol. Rhythms (2017) 15:159–166

modest. Therefore, additional measures to supplement and has antioxidant properties [33]. However, as the study by
revise current insomnia treatment are warranted to obtain Lin, Tsai [27] did not include a control group, it is difficult
enhanced treatment effects [13]. to conclude causally that kiwi has sleep improving prop-
One line of research that has received little attention erties. Thus, to clarify the effects of kiwi consumption on
up until now, but that could potentially improve current sleep, randomized controlled trials are needed.
insomnia treatment, is the relationship between sleep and Therefore, we designed the study with the purpose
nutrition. The limited amount of prior research suggests of replicating the aforementioned study with a control
a transactional relationship between these two processes. group to investigate whether kiwi has beneficial proper-
For instance, sleep deprivation has been found to increase ties on sleep. We conducted a randomized, controlled trial
levels of appetite regulating hormones, and reduced sleep addressing whether intake of kiwi fruit would improve
duration has accordingly been associated with metabolic sleep parameters in a student population suffering from
disorders and obesity [14, 15]. Reversely, it has been chronic insomnia symptoms. We hypothesized that the kiwi
shown, for example, that weight restoration in anorectic condition would yield better effects on objective and sub-
patients may improve subjective sleep quality [16]. jective sleep parameters as measured by actigraphy, sleep
Traditionally, nutritional research has been conducted by diary, and sleep questionnaires compared to the pear condi-
analyzing the effects of micro- and macro-nutrients, as well tion from baseline to post-treatment due to differences in
as whole foods, respectively. Micro-nutrients that have been relevant nutritional content.
investigated in conjunction with sleep include tryptophan
[17], vitamin B6 [18], vitamin B12 [19], vitamin D [20,
21], iron [22], and magnesium [23]. While this approach is Materials and methods
useful for understanding the mechanism of action of single
nutrients, its real-world application is often somewhat lim- Sample
ited as these supplements have to be extracted from other
food sources or produced in the laboratory. Normally, they In all, 74 students at the University of Bergen were
also need to be administered as pills. recruited (18 males and 56 females, mean age 24.3, SD
Research on nutrition and sleep has also studied the 3.2). The inclusion criteria used were identical with the
effects of macro-nutrients that comprise main nutritional fulfillment of the criteria for insomnia as measured by
components (i.e. protein, fats, and carbohydrates), as well the Bergen Insomnia Scale (BIS) [34]. Participants had to
as whole foods. In terms of macro-nutrients, some studies report nocturnal symptoms of insomnia (difficulties initiat-
have investigated the effects of protein and carbohydrate ing sleep, maintaining sleep, early morning awakenings, or
intake on sleepiness [24]. The effect on sleep of whole non-restorative sleep) at least 3 days per week (item 1–4 on
foods, such as fish, has also been investigated. In a rand- the BIS) as well as daytime consequences (sleepy/tiredness
omized controlled trial, we recently demonstrated that con- during daytime or being dissatisfied with sleep) at least
sumption of salmon had a small, albeit positive effect on 3 days per week (item 5 and 6 on the BIS) during the last
several sleep parameters [25]. Despite these findings, very month.
little is known about nutritional advices that can be sub- G*Power (version 3.1.3) was used for power calcula-
sumed under sleep hygiene recommendations [26]. tions [35]. The previous uncontrolled study by Lin and
Recently, in a free-living, self-controlled diet design, it Tsai [27] reported effects on some of the outcome meas-
was found that consuming two kiwi fruits prior to bedtime ures equaling effect sizes above 1.5 (Cohen’s d). In terms of
resulted in significant and large improvements on subjec- power calculations, we still decided to adhere to a relative
tive and objective measures of total sleep time and sleep conservative approach and based our power analysis on a
efficiency during a 4-week intervention [27]. It was hypoth- moderate effect size (Cohen’s d = 0.50). Alpha was set to
esized that this effect might have been related to the high 0.05 (two-tailed) and power was set to 80, and the correla-
levels of antioxidants, flavonoids, carotenoids, and antho- tion coefficient between the repeated measures was set to
cyanins found in kiwi. Oxidative stress seems to charac- 0.50. This suggested that we needed a total of 35 subjects
terize many patients with sleep disorders, and the afore- to detect significant group × time effects.
mentioned substances found in kiwi fruits are assumed to
improve sleep as they decrease oxidative stress [28]. Kiwi Procedure
is also rich in folate, low levels of which are associated
with insomnia [29, 30]. Furthermore, kiwi contains mela- Participants were recruited from university Websites. The
tonin [31] which plays an important role in terms of cir- advertisement included general information about the
cadian rhythm regulation [1]. Hypnotic effects of this hor- inclusion criteria, the length of the study, what partici-
mone have also been reported [32]; in addition, melatonin pants were required to do, anonymity, ethical approval,

13
Sleep Biol. Rhythms (2017) 15:159–166 161

compensation, and rationale. The provided rationale was Measures

to investigate the effects of different forms of nutrition
on sleep. The information also included a link to a Web- Demography  Questions about age, gender, living arrange-
based survey including demographics and the Bergen ment, study program, and part-time work were asked. These
Insomnia Scale. Those eligible for the study were invited questions were administered only at baseline.
to participate. Informed consent was obtained from all
individual participants included in the study. Partici- Sleep diary  A sleep diary comprising daily estimates of
pants were instructed to wear an actigraph on the non- the following sleep variables was kept for 1 week at baseline
dominant wrist during a baseline week, as well as during and during the last week of the intervention: (1) sleep onset
the final week of a 4-week long fruit intervention period. latency; (2) wake after sleep onset; (3) total sleep time; (4)
During baseline, the participants were further instructed sleep efficiency; (5) perceived sleep quality [ rated via a
to keep a sleep diary for 1 week. Participants also kept a Likert scale ranging from 1 (worst) to 5 (best)]; and (6) per-
sleep diary during the last week of treatment (post-treat- ceived daytime functioning [rated via a Likert scale ranging
ment). Moreover, the participants were randomized to eat from 1 (worst) to 5 (best)] [38].
either ~130  g of kiwi, the active condition, or pear, the
control condition, 1  h before bed time for 4  weeks fol- Actigraphy  Actigraphy consists of small wrist-worn units
lowing the baseline week. Intake of pear was decided to about the size of a watch and containing an accelerometer, a
be the control condition as it differs significantly from clock, and an automatic storing feature. Accordingly, wrist
kiwi as regards hypothesized sleep-conducive attributes. movements are recorded and stored. Recorded data are
For instance, kiwis contain on average 42  µg/100  g of transferred to a computer. Based on a software algorithm,
folate and 59  mg/100  g of Vitamin C, as compared to the data are analyzed to ascertain of sleep parameters [39].
7  µg/100  g and 4  mg/100  g in pear, respectively (Nor- In the present study, we used the Actiwatch 2 from Philips/
wegian Food Composition Database 2014. Norwegian Respironics (http://www.usa.philips.com/healthcare/prod-
Food Safety Authority. The Norwegian Directorate of uct/HC1044809/actiwatch-2-activity-monitor). Epoch time
Health and University of Oslo http://www.matvaretabel- was set to 1.00  min. Sensitivity was set to medium. Acti-
len.no). Furthermore, kiwi has been found to contain graphic data (total sleep time, wake after sleep onset, sleep
approximately 24 µg/g of melatonin, which has not been efficiency, and sleep onset latency) were collected at base-
detected in pear [31, 36]. Finally, the total average anti- line and at post-treatment, and were used as outcome vari-
oxidant content in kiwi is twice as high as pear [37]. Par- ables.
ticipants received money to buy the fruit required for the
study. They were also required to fill in a log to document Bergen Insomnia Scale  This scale consists of six items
adherence to fruit intake during the treatment period. adhering to criteria for insomnia stated in the Diagnostic
The project received ethical approval from the and Statistical Manual for Mental Disorders IV [40] and
Regional Committee for Medical and Health Research conforming to clinical criteria for defining insomnia [34]:
Ethics in Western Norway (No. 2014/2174)/REK West). during the last month, how many days per week (0–7): (1)
All participants were paid 500 NOK (~60  US$) for par- did you spend more than 30 min to fall asleep after turning
ticipating in addition to having their fruit expenses cov- out the lights? (2) Did you wake up for more than 30 min
ered. Data collection took place between March and during sleep? (3) Did you wake up more than 30 min earlier
May 2015. Randomization was performed electronically than you wanted, without being able to fall asleep again?
(http://www.randomizer.org). To blind the researcher (4) Have you not felt adequately rested following sleep?
administering the treatment (ØON), participants were (5) Have you been so sleepy/tired that it has interfered with
given a note identifying the condition to which they were school/job or your private life? (6) Have you been dissatis-
randomized; the note was placed in a sealed envelope, fied with your sleep? A total score ranging from 0 to 42 was
which they were instructed to open after the baseline calculated by adding the score of the six items [34]. The
week. Seven participants withdrew during the study due Cronbach alpha of the BIS at baseline and post-treatment
to time issues and travels, yielding a total number of 67 was 0.73 and 0.77, respectively. The BIS was also used for
respondents. However, there were missing data on some categorizing the respondents either as insomniacs or non-
of the measures, yielding actigraphy data for 65 partici- insomniacs post-treatment.
pants, sleep diary data for 53 participants, and survey
data for 66 participants, respectively. Reasons include Pittsburgh Sleep Quality Index  This instrument assesses
technical difficulties with some of the actigraphy record- quality of sleep and identifies sleep problems during the
ings, and lost or incomplete sleep diaries (see Fig. 1 for last month [41]. The questionnaire consists of 19 self-rated
consort flow diagram). questions yielding 7 components: subjective sleep quality,

13

162 Sleep Biol. Rhythms (2017) 15:159–166

Enrollment Assessed for eligibility (n= 74)

Excluded (n= 0)

Randomized (n= 74)

Allocation
Allocated to Kiwi intervention (n= 37) Allocated to Pear intervention (n= 37)

Follow-Up
Lost to follow-up (time issues/travels) (n= 6) Lost to follow-up (time issues/travels) (n= 1)

Analysis
Analysed Sleep Diary (n = 24) Analysed Sleep Diary (n = 29)
♦ Excluded from analysis (incomplete data) ♦ Excluded from analysis (incomplete data)
(n= 7) (n= 7)

Analysed Actigraphy (n= 30) Analysed Actigraphy (n= 35)

♦ Excluded from analysis (technical difficulties) ♦ Excluded from analysis (technical difficulties)
(n= 1) (n= 1 )

Analysed PSQI/BIS (n = 31) Analysed PSQI/BIS (n = 36)

Fig. 1  Consort flow diagram

sleep latency, sleep duration, sleep efficiency, sleep distur- administered. Cut-off for adherence was set to a minimum
bance, use of sleep medications, and daytime dysfunction. of 20 out of 28 days.
Each component is scored from 0 to 3, yielding a global
PSQI score between 0 and 21. The PSQI global score is Hypothesis Assumption Questionnaire  A questionnaire
the most commonly used parameter from the PSQI and was presented at post-treatment asking about participants’
was consequently used as an outcome variable in the pre-assumptions concerning of the hypothesis in the study
present study. The PSQI is validated in Norwegian with to rule out any influence of demand characteristics. Two
satisfactory validity and reliability [42]. At baseline and items were included: (1) do you have any assumptions
post-treatment, the Cronbach alpha for the PSQI was 0.54 regarding the hypothesis in this study (that is, whether one
and 0.55, respectively. of the nutrients are assumed to have a greater effect than the
other, or whether it is assumed that they should yield simi-
Adherence Log  A log for fruit intake where participants lar effects)? The participants could answer “yes” or “no”. If
checked off “yes” or “no” for each day of intervention was “yes”: (2) I think the hypothesis is that (i) kiwi works better

13
Sleep Biol. Rhythms (2017) 15:159–166 163

for insomnia than pear; (ii) pear works better for insomnia Sleep diary variables
than kiwi; (iii) there are no differences in effect on insomnia
between kiwi and pear. The results concerning sleep diary data are presented in
Table  1. None of the main effects of group were signifi-
cant, implying that overall across baseline and post-treat-
Statistics ment, there were not any differences on sleep diary param-
eters across the two conditions (kiwi vs. pear). There was
Data were analyzed with a 2  × 2 ANOVA with one a significant main effect of time on all subjective sleep
between-subject factor (group: kiwi vs. pear) and one measures except for total sleep time (TST), suggesting an
within-subject factor (time: baseline and post-treatment). overall improvement in subjective sleep from baseline to
Differences between conditions in terms of categorization post-treatment across conditions. There was a significant
as insomniacs were investigated by Chi-square analysis. group × time interaction effect on sleep quality and daytime
Significance level was defined as p < 0.05. functioning, showing that the improvement from baseline
to post-treatment was larger for the kiwi condition com-
pared to the pear condition. The corresponding effect size
(Cohens d) reflecting the relative changes across conditions
Results was 0.68 for sleep quality and 0.51 for daytime functioning,
respectively.
Demographics
Actigraphy data.
Participants in the kiwi (8♂, 23♀) and the pear condition
(5♂, 31♀) neither differed in terms of gender (χ2 = 0.219, The results regarding actigraphy data are presented in
df = 1, p > 0.05) nor in age (kiwi: m = 24.5, SD 3.6 vs. pear: Table 2. There was no significant main effect of group and
m = 24.1, SD 2.8; t = 0.39, df = 65, p = 0.69). time and no significant group × time interaction effect.

Table 1  Effects of kiwi and pear on sleep diary parameters

Variable Kiwi (n = 24) Pear (n = 29) Main effect group Main effect time Group × time interaction
Baseline Post Baseline Post
M SD M SD M SD M SD

Sleep onset latency (min) 44 23 29 17 42 24 29 20 F1,51 = 0.04,NS F1,51 = 16.32,** F1,51 = 0.11,NS

Wake after sleep onset (min) 17.8 14.4 11.1 11.6 17.9 18.6 12.9 17.2 F1,51 = 0.58,NS F1,51 = 13.84,** F1,51 = 0.31,NS
Total sleep time (min) 418 47 453 47 426 57 435 29 F1,51 = 0.12,NS F1,51 = 3.40,NS F1,51 = 0.26,NS
Sleep efficiency (%) 81.4 6.3 85.5 6.1 83.0 7.7 86.4 7.8 F1,51 = 0.53,NS F1,51 = 15.53,** F1,51 = 0.15,NS
Sleep quality (1–5) 2.9 0.4 3.4 0.4 3.1 0.6 3.2 0.6 F1,51 = 0.40,NS F1,51 = 19.33,** F1,51 = 5.88,*
Daytime functioning (1–5) 2.4 0.5 3.0 0.6 2.6 0.5 2.8 0.6 F1,51 = 0.06,NS F1,51 = 27.29,** F1,51 = 4.79,*

ns not significant
*p < 0.05, **p < 0.01

Table 2  Effects of kiwi and pear on actigraphic parameters

Variable Kiwi (n = 30) Pear (n = 35) Main effect group Main effect time Group × time interaction
Baseline Post Baseline Post
M SD M SD M SD M SD

Total sleep time (min) 395 36 396 37 407 50 409 52 F1,63 = 1.43,NS F1,63 = 0.11,NS F1,63 = 0.00,NS
Wake after sleep onset (min) 52.3 18.5 53.8 19.3 50.7 15.5 50.9 18.7 F1,63 = 0.29,NS F1,63 = 0.20,NS F1,63 = 0.12,NS
Sleep efficiency (%) 82.2 4.4 82.6 4.7 83.6 5.3 83.6 5.2 F1,63 = 0.99,NS F1,63 = 0.12,NS F1,63 = 0.21,NS
Sleep onset latency (min) 22 15 18 10 17 14 18 12 F1,63 = 0.44,NS F1,63 = 1.29,NS F1,63 = 2.38,NS

ns not significant

13

164 Sleep Biol. Rhythms (2017) 15:159–166

Sleep questionnaires Discussion

The results of the PSQI and BIS questionnaires are pre- In this randomized controlled trial, we sought to ascertain
sented in Table 3. None of the main effects of group were any differential effects of kiwi- and pear consumption on
significant. There was a significant main effect of time sleep parameters in a student population reporting chronic
across both conditions on both the PSQI and the BIS, insomnia symptoms. Of particular interest was any interac-
reflecting a general improvement in sleep from baseline tion effects showing differences over time between the con-
to post-treatment. No significant group  ×  time interaction ditions. In terms of the sleep diary data, there was no main
effect was found. In all, 25.8 and 30.6% in the kiwi and effect of group, but a significant effect of time was found
the pear condition, respectively, were categorized as non- on five (sleep latency, wake after sleep onset, sleep effi-
insomniacs at post-treatment (χ2  =  0.02, df = 1, p > 0.05, ciency, sleep quality, and daytime functioning) of six sleep
continuity corrected). diary outcome variables, all suggesting better sleep at post-
treatment compared to baseline. On two (sleep quality and
Adherence data daytime functioning) of six sleep diary outcome variables,
significant interaction effects were found, suggesting more
All completers complied with the required fruit intake. improvement in the kiwi compared to the pear condition.
Mean number of days (out of 28) of fruit intake was 26.0, The actigraphic data showed no a main effect of group or
SD 1.9, specifically 26.2 (SD 1.7) in the kiwi and 25.8 (SD time, nor any group × time interaction effects. On the two
2.1) in the pear condition. sleep questionnaires (the Pittsburgh Sleep Quality Index
and the Bergen Insomnia Scale), a main effect of time was
Assumption about the hypothesis at post‑treatment found, reflecting improvement from baseline to post-treat-
ment; however, no main effect of group and no significant
To ensure that demand characteristics [43] had not influ- group  ×  time interaction effect were detected. In terms of
enced the results, we asked all participants, as recom- categorization as insomniacs, no difference between the
mended by Rubin and Paolini [44], at post-treatment if conditions was detected.
they had any assumption about which intervention (kiwi vs. Overall, the findings based on the subjective sleep
pear) would have the best effect. data (sleep diary and questionnaires) suggest a significant
In all, 71.4% (of which 45.2% from the kiwi and 54.8% improvement from baseline to post-treatment, irrespective
from the pear condition) did not have any predictions, of intervention. Although most studies suggest that sponta-
11.6% predicted that kiwi would have a greater effect than neous recovery seldom occurs in insomniacs [45], placebo
pear (of which 57.1% kiwi and 42.9% pear condition), and effects have been noticed in several randomized insomnia
8.5% thought pear would be superior to kiwi (of which treatment trials [46]. The general improvement reported by
60.0% kiwi and 40.0% pear condition). Finally, 8.5% (of the participants across conditions in the present study may
which 80.0% kiwi and 20.0% pear condition) assumed thus be attributed to placebo effects, and strongly empha-
that there would be no differential effects between the two sizes the need for proper experimental control conditions,
conditions. which unfortunately was lacking in the kiwi-intervention
study by Lin, Tsai [27]. The fact that no significant effects
were found on any of the objective data underlines that
insomnia mainly is a subjective disorder, often character-
ized by discrepancies between subjective and objective
sleep recordings [47]. Lack of findings on the actigraphic

Table 3  The effects of kiwi and pear on the Pittsburgh Sleep Quality Index (PSQI) and the Bergen Insomnia Scale (BIS)
Variable Kiwi (n = 31) Pear (n = 36) Main effect group Main effect time Group × time interaction
Baseline Post Baseline Post
M SD M SD M SD M SD

PSQI—global score 10.8 2.4 7.8 2.7 10.8 2.9 7.5 2.3 F1,64 = 0.14,NS F1,64 = 83.95,** F1,64 = 0.17,NS
BIS—total score 21.9 6.6 15.0 6.3 21.6 6.3 15.7 7.2 F1,64 = 0.01,NS F1,64 = 70.87,** F1,64 = 0.50,NS

ns not significant
**p < 0.01

13
Sleep Biol. Rhythms (2017) 15:159–166 165

data may also reflect that subjective data are more sensitive recovery as well as placebo effects. Another asset of the
to changes in insomnia state than objective data [48]. present study was the use of both subjective and objective
In terms of interaction effects, the present study showed sleep measures. Including measures of both fruit intake
that on two sleep diary parameters (sleep quality and day- adherence and the participants’ predictions of the study’s
time functioning), the kiwi group fared better than the pear hypothesis contributed further to methodological rigor.
group. Thus, it cannot be ruled out that the kiwi fruit may In summary, on a total of two (sleep quality and daytime
have some sleep-conducive properties, possibly due to sub- functioning) out of 12 outcome variables, a statistically sig-
stances such as antioxidants, flavonoids, carotenoids, and nificant group  ×  time interaction effect favoring the kiwi
anthocyanins [28], folate [29], and melatonin [31]. Interest- condition was found. Based on this study, one cannot rule
ingly, the trial by Lin and Tsai [27] did not include the two out the possibility that kiwi may have some sleep improv-
outcome measures (sleep quality and daytime functioning) ing properties. However, this needs to be confirmed by
that turned out to be significant in the present study. Con- future studies.
versely, several of the outcome measures that Lin and Tsai
Compliance with ethical standards 
[27] reported as significant (such as sleep onset latency,
wake after sleep onset, and total sleep time) were not sig-
Ethical approval  Regional Committee for Medical and Health
nificant (group × time) in the present study. More well-exe- Research Ethics in Western Norway No. 2014/2174/REK West.
cuted trials, including subjective as well as objective data,
are needed to conclude if kiwi has any sleep-facilitating Conflict of interest  None of the authors declares any conflicts of
effects. interest.
Some limitations of the present study should be noted.
The current study was of short duration and did not have Funding  University of Bergen and the Regional Research Council of
Norway provided funding.
any follow-up measurements that could reveal potential
long-term effects. Although the study included actigraphy
as an objective measure, the gold standard, polysomnogra-
phy [49], was not included. This prevented us from inves-
tigating effects in terms of sleep architecture. There were References
some missing data in the data set. However, in all cases, the
1. Morin CM, Benca R. Chronic insomnia. Lancet.
number of observations retained for analyses was sufficient 2012;379(9821):1129–41.
according to the a priori power analysis. The sample con- 2. American Psychiatric Association. Diagnostic and statistical
sisted exclusively of students, limiting the generalizabil- manual of mental disorders, 5th ed. 2013. American Psychiatric
Association, Washington, DC.
ity of the findings. Nevertheless, all participants reported
3. Ohayon MM. Epidemiology of insomnia: what we know and
chronic insomnia symptoms at treatment initiation. As no what we still need to learn. Sleep Med Rev. 2002;6(2):97–111.
formal clinical interviews or polysomnographic screenings 4. Pallesen S, et al. Prevalence of insomnia in the adult Norwegian
were conducted, we cannot rule out that insomnia symp- population. Sleep. 2001;24(7):771–9.
5. Pallesen S, et al. A 10-year trend of insomnia prevalence in the
toms in some cases might have reflected sleep problems
adult Norwegian population. Sleep Med. 2014;15:173–9.
such as delayed sleep phase syndrome and obstructive 6. Sivertsen B, et al. Does insomnia predict sick leave? The Horda-
sleep apnea. Although pear did contain some of the same land Health Study. J Psychosom Res. 2009;66:67–74.
relevant micro-nutrients as kiwi, the concentration of these 7. Sivertsen B, et al. The long-term effect of insomnia on work dis-
ability—the HUNT-2 historical cohort study. Am J Epidemiol.
was far higher in the latter, which arguably makes pear
2006;163:1018–24.
relevant as a control condition. In terms of limitations, it 8. Baglioni C, et al. Insomnia as a predictor of depression: a meta-
should also be noted that the Cronbach alpha of the PSQI analytic evaluation of longitudinal epidemiological studies. J
was relatively low. Affect Disord. 2011;135:10–9.
9. Sivertsen B, et  al., Midlife insomnia and subsequent mortality:
Some notable strengths of the present study also deserve
the Hordaland health study. BMC Public Health, 2014. 14:article
mention. The participants were not informed about which no. 720.
condition was assumed to be superior, as the study admin- 10. Rybarczyk B, et al. Cognitive behavioral therapy for insomnia in
istrator was blind to condition. Still, seven participants older adults: background, evidence, and overview of treatment
protocol. Clin Gerontol. 2013;36:70–93.
were able to correctly guess the hypothesis at the end of
11. Qaseem A, et  al. Management of chronic insomnia disorder in
the study. This amounts to only 10.4% of the participants, adults: a clinical practice guideline from the American College
making it improbable that demand characteristics influ- of Physicians Management of Chronic Insomnia Disorder in
enced the findings [43]. The fact that the study administra- Adults. Ann Intern Med. 2016;165:125–33.  
12. Edinger JD, Means MK. Cognitive-behavioral therapy for pri-
tor was blind to condition prevented the results from being
mary insomnia. Clin Psychol Rev. 2005;25(5):539–58.
influenced by experimental bias [50]. The inclusion of a 13. Harvey AG, NKY Tang. Cognitive behaviour therapy for primary
control group made it possible to control for spontaneous insomnia: can we rest yet? Sleep Med Rev. 2003;7(3):237–62.

13

166 Sleep Biol. Rhythms (2017) 15:159–166

14. Morselli L, et al. Role of sleep duration in the regulation of glu- 33. Bonnefont-Rousselot D, Collin F, Melatonin. Action as anti-

cose metabolism and appetite. Best Pract Res Clin Endocrinol oxidant and potential applications in human disease and aging.
Metab. 2010;24(5):687–702. Toxicology. 2010;278(1):55–67.
15. Spiegel K, et  al., Sleep loss: a novel risk factor for insulin
34. Pallesen S, et al. A new scale for measuring insomnia: the Ber-
resistance and Type 2 diabetes. J Appl Physiol (1985), 2005. gen Insomnia Scale. Percept Mot Skills. 2008;107(3):691–706.
99(5):2008–19. 35. Faul F, et al. G*Power 3: a flexible statistical power analysis pro-
16. Pieters G, et  al. Sleep variables in anorexia nervosa: evolution gram for the social, behavioral, and biomedical sciences. Behav
with weight restoration. Int J Eat Disord. 2004;35(3):342–7. Res Methods. 2007;39(2):175–91.
17. Peuhkuri K, Sihvola N, Korpela R. Diet promotes sleep duration 36. Huang X, Mazza G. Simultaneous analysis of serotonin,

and quality. Nutr Res. 2012;32:309–19. melatonin, piceid and resveratrol in fruits using liquid chro-
18. Ebben M, Lequerica A, Spielman A. Effects of pyridox-
matography tandem mass spectrometry. J Chromatogr A.
ine on dreaming: a preliminary study. Percept Mot Skills. 2011;1218(25):3890–9.
2002;94:135–40. 37. Szeto YT, Tomlinson B, Benzie IF. Total antioxidant and

19. Ohta T, et al. Treatment of persistent sleep-wake shcedule disor- ascorbic acid content of fresh fruits and vegetables: implica-
ders in adolescents with methylcobalamin (vitamin-B12). Sleep. tions for dietary planning and food preservation. Br J Nutr.
1991;14:414–8. 2002;87(1):55–9.
20. Gominak SC, Stumpf WE. The world epidemic of sleep dis- 38. Carney CE, et al. The consensus sleep diary: standardizing pro-
orders is linked to vitamin D deficiency. Med Hypotheses. spective sleep self-monitoring. Sleep. 2012;35(2):287–302.
2012;79(2):132–5. 39. Sadeh A. The role and validity of actigraphy in sleep medicine:
21. Huang W, et al. Improvement of pain, sleep, and quality of life an update. Sleep Med Rev. 2011;15(4):259–67.
in chronic pain patients with vitamin D supplementation. Clin J 40. American Psychiatric Association. Diagnostic and statistical

Pain. 2013;29(4):341–7. manual of mental disorders, 4th ed. 1994. American Psychiatric
22. Kordas K, et al. The effects of iron and/or zinc supplementation Association, Washington, DC.
on maternal reports of sleep in infants from Nepal and Zanzibar. 41. Buysse DJ, et  al. The Pittsburgh Sleep Quality Index: a new
J Dev Behav Pediatr. 2009;30:131–9. instrument for psychiatric practice and research. Psychiatry Res.
23. Held K, et al. Oral Mg(2+) supplementation reverses age-related 1989;28(2):193–213.
neuroendocrine and sleep EEG changes in humans. Pharmacops- 42. Pallesen S, et  al. Pittsburgh sleep quality index. Tidsskrift for
ychiatry. 2002;35:135–43. Norsk Psykologforening. 2005;42(8):714.
24. Spring B, et  al. Effects of protein and carbohydrate meals on 43. Orne MT. On the social psychology of the psychological experi-
mood and performance—interactions with sex and age. J Psychi- ment: with particular reference to demand characteristics and
atr Res. 1983;17:155–67. their implications. Am Psychol. 1962;17(11):776.
25. Hansen AL, et  al. Fish consumption, sleep, daily functioning, 44. Rubin M, Paolini S, Crisp RJ. A processing fluency explanation
and heart rate variability. J Clin Sleep Med. 2014;10(5):567–75. of bias against migrants. J Exp Soc Psychol. 2010;46(1):21–8.
26. Stepanski EJ, Wyatt JK. Use of sleep hygiene in the treatment of 45. Matteson-Rusby SE, et  al. Why treat insomnia? Primary Care
insomnia. Sleep Med Rev. 2003;7(3):215–25. Companion J Clin Psychiatry. 2010;12(1):PCC.08r00743.
27. Lin HH, et  al. Effect of kiwifruit consumption on sleep qual- 46. Perlis ML, et al. Placebo effects in primary insomnia. Sleep Med
ity in adults with sleep problems. Asia Pac J Clin Nutr. Rev. 2005;9(5):381–9.
2011;20(2):169–74. 47. Voderholzer U, et  al. Are there gender differences in objec-

28. Tsaluchidu S, et al. Fatty acids and oxidative stress in psychiatric tive and subjective sleep measures? A study of insomniacs and
disorders. BMC Psychiatry. 2008;8(1):1–3. healthy controls. Depress Anxiety. 2003;17(3):162–72.
29. Indrawati et al. Comparative study on pressure and temperature 48. Lund HG, et al. The discrepancy between subjective and objec-
stability of 5-methyltetrahydrofolic acid in model systems and in tive measures of sleep in older adults receiving CBT for comor-
food products. J Agric Food Chem. 2004;52(3):485–92. bid insomnia. J Clin Psychol. 2013;69(10):1108–20.
30. Kelly GS, Folates. supplemental forms and therapeutic applica- 49. Carskadon MA, Dement WC. Normal human sleep: an overview.
tions. Altern Med Rev. 1998;3(3):208–20. Principles Pract Sleep Med. 2005;4:13–23.
31. Hattori A, et  al. Identification of melatonin in plants and
50. Rosenthal R. Covert communication in the psychological experi-
its effects on plasma melatonin levels and binding to mela- ment. Psychol Bull. 1967;67(5):356.
tonin receptors in vertebrates. Biochem Mol Biol Int.
1995;35(3):627–34.
32. Buscemi N, et  al. The efficacy and safety of exogenous mela-
tonin for primary sleep disorders. A meta-analysis. J Gen Intern
Med. 2005;20(12):1151–8.

13