RESEARCH

doi: 10.1111/nicc.12428

Back massage intervention for improving

health and sleep quality among intensive
care unit patients
Wen-Chi Hsu∗ , Su-Er Guo∗ and Chia-Hao Chang

ABSTRACT
Background: A massage may relax muscles, improve blood circulation and reduce pain and anxiety while also improving sleep quality by increasing comfort. However, there is little
research on whether a back massage improves sleep quality in intensive care unit (ICU) patients.
Aims and objectives: This study examined the effects of a back massage on improving vital signs, sleep quality, anxiety and depression among ICU patients.
Design: Adopting a quasi-experimental design, convenience sampling was used to recruit ICU patients from a medical centre in Southern Taiwan. The experimental group received
back massages for three consecutive days (n = 30), while controls received usual care (n = 30).
Methods: The Verran and Snyder-Halpern Scale and the Hospital Anxiety and Depression Scale were used, and subjective and objective sleep time (wrist actigraphy and sleep
duration from nurse observations) was recorded. The effect of the intervention was examined using a generalized estimating equation model with a robust standard error and an
exchangeable working correlation matrix adjusting for time.
Results: The results show that subjective sleep quality scores in ICU patients were low. Mean observed sleep time (measured by nurses) was 3⋅9 h, but mean sleep time measured
using wrist actigraphy was 5⋅9 h. Back massages improved breathing in patients, increased sleep quality reflected by both subjective and objective data and were associated with a
significant change in anxiety.
Conclusions: These findings suggest that a 10-min back massage can improve sleep quality, sleep duration, breathing and anxiety in ICU patients.
Relevance to clinical practice: The implementation of a back massage shows positive improvements in the sleep quality of ICU patients. The training and theory of massage
interventions should be further applied when developing courses in critical care nursing.
Key words: Actigraphy • Anxiety • Critically ill patients • Massage • Sleep quality

INTRODUCTION
Sleep is vital for the maintenance of an indi-
vidual’s physical and psychological health.
There are several syndromes related to poor
sleep, including fatigue, daytime sleepiness,
depression and anxiety. Sleep disorders can
cause over-excitability of the sympathetic
nervous system while decreasing parasym-
pathetic nervous system activity. Moreover,
sleep disorders increase pain sensitivity,
agitation, nervous tension, tachycardia
and irregular heart rate (HR). They can also
have an adverse impact on the immune
system, metabolism and nervous system
and lead to cognitive dysfunction, as well
as an increased risk of depression (Li et al.,
2016; Salo et al., 2012).
BACKGROUND
Approximately a third (24⋅3–44⋅1%)
of adults in the USA have reported dis-
turbed or poor sleep (Centers for Disease
Control and Prevention, 2017). Sleep dis-
ruption is more common in the intensive
care unit (ICU) population than in healthy
adults (Elliott et al., 2013). Studies have
shown that ICU patients experience poor
sleep quality (Shiung et al., 2009; Pisani
et al., 2015), possibly because of frequent
treatment, continuous light exposure, noise
from equipment and ambient noise in an
emergency, mechanical ventilation, medica-
tion side effects and their disease conditions
(Boyko et al., 2012). Insufficient sleep among
critically ill patients can have adverse
catabolic effects and negatively impact
recovery, thus extending hospitalization
durations and increasing medical costs (Xu
et al., 2016; Choi et al., 2017). Poor sleep qual-
ity has been shown to be associated with a
higher likelihood of health care use (Choi

∗ Wen-Chi Hsu and Su-Er Guo contributed equally to this work as first authors.
Authors: W-C Hsu, MSN, RN, Clinical Nurse Specialist, Chi Mei Medical Center, Tainan, Taiwan; S-E Guo, PhD, RN, Professor and Director, Graduate Institute of Nursing,
College of Nursing, Chang Gung University of Science and Technology (CGUST), Puzi, Taiwan; Chronic Diseases and Health Promotion Research Centre, CGUST, Puzi, Taiwan; Division
of Pulmonary and Critical Care Medicine, Chiayi Chang Gung Memorial Hospital, Chang Gung Medical Foundation, Puzi, Taiwan; Department of Safety Health and Environmental
Engineering, Ming Chi University of Technology, New Taipei City, Taiwan; C-H Chang, PhD, Associate professor, Graduate Institute of Nursing, College of Nursing, CGUST, Puzi,
Taiwan
Address for correspondence: Su-Er Guo, Director, Graduate Institute of Nursing, College of Nursing, Chang Gung University of Science and Technology, Puzi City, Chiayi
County, Taiwan, 2, Sec. W., Jiapu Road, Puzi City, Chiayi County 61363, Taiwan
E-mail: seguo@mail.cgust.edu.tw, sxg90huang@gmail.com

© 2019 British Association of Critical Care Nurses 1

Back massage and sleep quality
et al., 2017). Moreover, the impact of poor
sleep quality on health appears to remain
after discharge from the ICU (Altman et al.,
2017).
There are several interventions used to
improve sleep quality, including (1) relax-
ation training, such as progressive muscle
relaxation and meditation; (2) decreas-
ing environmental noise or light; and (3)
acupressure, music therapy or a combina-
tion therapy. Massages can relax muscles,
improve blood circulation, reduce pain
and anxiety and improve sleep quality
by increasing comfort (Fang and Wang,
2007; Richards et al., 2003). Although back
massages promote recovery via enhanced
sleep and immune function and have been
approved for use in an acute care setting
(Lareau et al., 2008), they are rarely used
in the ICU. Only one study has investi-
gated the effectiveness of back massages
on sleep quality (Shinde and Anjum, 2014),
and this study predominantly examined
the effects of only a single intervention,
whereas the cumulative effect of multiple
interventions was not evaluated. In addi-
tion, the researchers did not evaluate vital
signs, anxiety and depression status, which
are important for recognizing the multiple
effects of a back massage and for validating
its future use.
AIMS AND OBJECTIVES
This study therefore examined the effects
of a back massage on sleep quality, anxiety
and depression among ICU patients. Our
findings can provide strategies for medical
professionals to improve sleep problems
among ICU patients, thereby further
reducing patients’ physical problems
and duration of hospitalization by improv-
ing the quality of nursing care.
DESIGN AND METHODS
Design and sample
This quasi-experimental study was per-
formed at the medical ICU of a medical
centre in Southern Taiwan, using pre-
and post-tests to examine the effects of back
massage on vital signs, sleep quality, anxiety
and depression. A convenience sample of 60
medical patients was recruited and allocated
to either the experimental group (n = 30) or
the control group (n = 30).
Patients were recruited if they were
(1) 18 years of age or over, (2) admitted
2 © 2019 British Association of Critical Care Nurses
to the ICU for over 48 h and had diffi-
culties falling asleep, (3) in a conscious
and stable condition and (4) were able
to communicate in oral/written Chinese
or Taiwanese. Patients were excluded if
they (1) were hard of hearing/had hear-
ing loss, (2) were blind/severely visually
impaired, (3) abused alcohol, (4) had
brain damage following surgery, (5) had
dementia/psychosis, (6) had a previous
back/lung/abdominal operation, (7) had a
skin wound on the back or (8) took sleeping
pills before admission to the ICU. Inclu-
sion and exclusion criteria were confirmed
through a thorough review of eligible
patient charts.
Procedures
Patients were randomly assigned to two
different groups using the opaque sealed
envelope method. The experimental group
(n = 30) received a 10-min back massage
at 9 PM on three consecutive nights (see
Table S1, Supporting Information for full
details), while the control group (n = 30)
received usual care without a back massage.
Assessments of vital signs, sleep, anxiety
and depression were conducted at baseline
and on each day of the intervention for
3 days (Figure S1).
Measures
Verran and Snyder-Halpern (VSH) Sleep
Scale: The VSH Sleep Scale, translated
and validated by Lin and Tsai (2003), was
administered to measure a patient’s mag-
nitude and self-reported quality of sleep
during the preceding night. It comprises
15 items: 8 items about sleep disturbance,
4 about effectiveness of sleep and 3 about
sleep compensation. Each item is rated on a
100-mm visual analogue scale (total scores
ranging from 0 to 1500). Lower scores rep-
resent worse quality of sleep. Cronbach’s
alpha coefficient, indicating reliability, was
0⋅82 in the original version (Snyder-Halpern
and Verran, 1987) and 0⋅83 in the Chinese
version (Lin and Tsai, 2003). In this study,
Cronbach’s alpha reliability coefficient was
0⋅80.
Wrist Actigraphy: Wrist actigraphy with
the Actiwatch 2 (PHILIPS Respironics,
Inc., Guildford, UK) was used to estimate
sleep–wake cycles. Data collected included
sleep onset latency, wake time after sleep
onset, total sleep time (TST) and sleep effi-
ciency. The Actiwatch was calibrated before
each use.
Sleep duration (nurse observation):
From 10 PM to 6 AM, primary nurses
observed and recorded the patients’ sleep-
ing condition every hour. Participants
were designated as (1) sleeping, (2) awake
or (3) unsure whether patient is asleep
or awake. Nurses also recorded the total
hours slept by 7 AM each morning. Nurses
classified a patient as sleeping when the
patient closed his/her eyes and fell asleep
in a short time and did not show any
motion thereafter. We attempted to min-
imize possible observation/information
bias by conducting a 1-h training course, in
which the principles of each classification
were explained and demonstrated to the
nurses. At the end of the training course, we
asked the nurses to rate the sleep duration
of 10 patients. We then calculated their
Cohen’s kappa coefficient using SPSS soft-
ware. The inter-rater reliability coefficient
was 0⋅91.
Hospital Anxiety and Depression Scale
(HADS): Developed by Zigmond and
Snaith (1983) and validated in a previ-
ous study (Turk et al., 2015), the HADS
assesses common clinical symptoms of anx-
iety and depression. It contains 14 items
on two subscales (7 items for depression
and anxiety each). Items are scored on a
Likert scale from 0 (completely disagree)
to 3 (strongly agree). Total scores for each
subscale were calculated, with higher total
subscale scores indicating greater anxiety
or depression (Zigmond and Snaith). Cron-
bach’s alpha reliability coefficient of the
Chinese version of the scale for anxiety
and depression was 0⋅75 and 0⋅76, respec-
tively (Yang et al., 2014). In the present
study, these values were 0⋅78 and 0⋅75,
respectively.
Frequency of Multiple-treatment Activ-
ity: A chart depicting the frequency of
multiple treatment was developed by our
research team, based on previous studies
(Delaney et al., 2015; Shiung et al., 2009), to
record the frequency of multiple-treatment
activity among patients in the ICU between
10 PM at night and 6 AM the next morning.
The items on the chart are the same as those
on the Perception of Multiple-Treatment
Interference Scale (see next section below).
Treatment activity includes several assess-
ments, which are typically scheduled on
an hourly basis, such as position change,
feeding, vital sign monitoring, fluid balance
assessment and pharmacological admin-
istration. Based on reviews of charts on
the frequency of implementing treatment,

if there were various medications with an
intravenous bolus applied at the same time,
only one would be calculated and included
in the chart.
Perception of Multiple-Treatment Inter-
ference Scale: This questionnaire is designed
for patients to evaluate the interference of
each treatment with sleep. The treatments
include blood tests; sputum suction; and
non-invasive treatments such as blood pres-
sure tests, a body temperature test, posi-
tion change, feeding, an alarm set for too
high or too low levels of blood pressure or
HR and so forth. This questionnaire uses
a 5-point Likert-type scale, in which each
item is scored from 1 (no interference) to
5 (extreme interference). The scores related
to a particular treatment were not calcu-
lated if the patient did not receive that treat-
ment. The scores were calculated based on
the total scores divided by the number of
questions answered. The content validity
index (CVI), defined as the proportion of
items on an instrument rated 3 or 4 by
all content experts (Polit and Beck, 2006),
was 0⋅86.
Vital Signs Monitoring: A Spacelabs Med-
ical (Washington, D.C., USA) monitor was
used to store and display 24-h physiological
parameters, with estimations of HR, respi-
ratory rate (RR) and mean arterial pressure.
These same parameters were estimated at
baseline and 10 min after the intervention on
each day. The monitor was calibrated before
each use.
Statistical methods
Data analyses were performed with SPSS
version 22.0 (IBM Corp., Armonk, NY, USA).
Frequency distributions, descriptive statis-
tics, chi-square tests and independent t-tests
were used. The effect of the intervention was
examined using a generalized estimating
equation (GEE) model with a robust stan-
dard error and an exchangeable working
correlation matrix adjusting for time (Liang
and Zeger, 1986). In this analysis, the inter-
vention effect would be supported if sig-
nificant two-way interaction effects (of time
by treatment) were found, indicating a sig-
nificant difference between the groups from
pre-test to post-test.
Ethical and research approvals
Approval was obtained from the hospital’s
institutional review board (IRB number
10212-011). Informed consent was obtained
from all participants.
© 2019 British Association of Critical Care Nurses
RESULTS
Participant characteristics
Sixty participants (mean age,
62⋅4 ± 11⋅8 years) were recruited (30 in each
group). Most participants were female, mar-
ried, unemployed and housewives and had
one or two chronic diseases. Most patients
had elementary school-level education. The
mean disease severity (APACHE II) was
15⋅47 ± 2⋅34, and most pain scores ranged
from 0 to 3. Age, gender, socio-economic
status, education level and relationship
status (married or single) have been
reported to affect sleep deprivation (Shinde
and Anjum, 2014). However, the demo-
graphic characteristics of the groups in the
present study did not differ significantly
(Table 1).
Participants’ health status and sleep
quality
Table 1 shows the mean and standard devi-
ations for HR, RR, mean arterial pressure,
subjective sleep (VSH Sleep Scale), objective
sleep (wrist actigraphy and nurse observa-
tions), anxiety and depression. Compared
with the control group, the experimental
group had worse subjective sleep quality
(p = 0⋅04). There were no significant dif-
ferences in any other measures (Table 1).
Specifically, there were no significant differ-
ences between the two groups with respect
to environmental and treatment variables.
Frequency (18⋅3 ± 4⋅8 versus 17⋅5 ± 3⋅0) and
level of multiple-treatment interference
(3⋅0 ± 0⋅6 versus 2⋅4 ± 0⋅6) in both groups
were similar and showed no significant
differences.
The mean score for subjective sleep qual-
ity in all ICU patients (experimental and
control group) was 648⋅3 (maximum total
score: 1500), reflecting poor sleep quality.
Moreover, the nurses observed an average
sleep duration of just 3⋅9 h, while wrist actig-
raphy recorded a sleep duration of 5⋅9 h.
Thus, both subjective and objective mea-
sures of sleep quality among ICU patients
in this study indicated poor sleep quality
(Table 1).
Effects of back massage on vital signs
GEE analysis indicated no difference
in change of HR and mean arterial pressure
from baseline between the experimen-
tal group and the control group at any
time point (Experimental group: first day
p = 0⋅80, second day p = 0⋅74 and third day
p = 0⋅96; Control group: first day p = 0⋅86,
Back massage and sleep quality
second day p = 0⋅72 and third day p = 0⋅63),
suggesting that back massage had no sig-
nificant effect on HR and mean arterial
pressure (Table 2).
In contrast, there was a significant dif-
ference between the experimental and con-
trol group regarding the change in RR from
baseline to the third day (p = 0⋅03), indicat-
ing that back massage had a significant effect
on breathing by the third day of the interven-
tion (Table 2, Figure S2A).
Effects of back massage on sleep
quality
GEE indicated no significant difference
in the improvement of subjective sleep
quality between the experimental and con-
trol groups from baseline to the first day
(p = 0⋅36). Nevertheless, there was a sig-
nificant difference between the groups in
improvement of subjective sleep quality
from baseline to the second and third days
(both p < 0⋅001), suggesting that back mas-
sage had a significant effect from the second
day of the intervention (Table 2, Figure S2B).
There was no significant difference in
the improvement of objective sleep quality
(wrist actigraphy) between the experimen-
tal and control groups from baseline to the
first day (p = 0⋅06). However, there was a
significant difference in improvement of
objective sleep quality between the groups
from baseline to the second and third days
(p = 0⋅04; p < 0⋅001, respectively; Table 2,
Figure S2C). Similar results were seen for
the other objective (nursing observations)
sleep quality measures, with no significant
differences in the improvement of objec-
tive sleep quality from baseline to the first
day (p = 0⋅53) but a significant difference
between the groups in the improvement of
objective sleep quality between baseline and
the second and third days (p = 0⋅04; p = 0⋅03,
respectively; Table 2, Figure S2D). Again,
these results indicate that back massage
had a significant effect from the second day
of intervention (Table 2). Overall, regard-
less of whether subjective or objective data
were used, all results indicate that back
massages influenced sleep quality from the
second day.
Effects of back massage on anxiety
and depression
In this sample, participants in the experi-
mental and control group had similar levels
of anxiety (12⋅1 versus 12⋅3) and similar bor-
derline levels of depression (9⋅7 versus 9⋅5).
3
Back massage and sleep quality

Table 1 Attributes, clinical characteristics and health status of patients in two groups (N = 60) the second longitudinal study that focuses
on examining the effects of back massage
Experimental group Control group on sleep quality and other health conditions
Characteristics Total, n (%) (n = 30), n (%) (n = 30), n (%) 𝜒2 p in ICU patients.
The average VSH Sleep Scale score,
Gender 0⋅63 0⋅42∗ self-reported by patients in the current
Male 23 (38⋅3) 14 (46⋅7) 9 (33⋅3) study, indicated poor sleep quality, which
Female 37 (61⋅7) 16 (53⋅3) 21 (66⋅7) was predicted and therefore not surprising.
Education level 1⋅89 0⋅59† Many studies confirmed that ICU patients
Elementary school 32 (53⋅4) 16 (53⋅3) 16 (53⋅3) frequently experience poor sleep, charac-
Middle and high school 22 (21⋅2) 12 (40) 10 (33⋅3) terized by prolonged sleep latency and
University and above 6 (13⋅3) 2 (6⋅7) 4 (13⋅3) frequent disruption (Bihari et al., 2012; Little
Religion 0⋅10 0⋅95† et al., 2012; Pisani et al., 2015). A possible
Taoism 39 (65) 20 (66⋅7) 19 (63⋅3) reason for the shorter duration or poorer
Buddhism 13 (24⋅5) 6 (20) 7 (23⋅3) quality of sleep in critically ill patients
Catholicism/Christianity 8 (15⋅1) 4 (13⋅3) 4 (13⋅3)
might be overexposure to light as the lights
Marital status 0⋅48 0⋅92†
in the ICU are constantly left on. Previous
Single 7 (11⋅7) 4 (13⋅3) 3 (10)
research suggests that light may reduce
Married 39 (65) 19 (63⋅3) 20 (66⋅7)
melatonin secretion, and thus sleep, and
Divorced 5 (8⋅3) 3 (10) 2 (6⋅7)
has effects on normal circadian rhythms,
Widowed 9 (15) 4 (13⋅3) 5 (16⋅7)
thereby increasing sleep latency but hin-
Occupation 1⋅96 0⋅58†
Housewife/Unemployment 33 (55) 16 (53⋅3) 17 (56⋅7) dering the optimization of sleep quality
Government employees 12 (20) 8 (26⋅7) 4 (13⋅3) (Gaggioni et al., 2014; LeGates et al., 2014;
Agriculture 5 (8⋅3) 2 (6⋅7) 3 (10) Shiung et al., 2009). Patients in the current
Labour 10 (16⋅7) 4 (13⋅3) 6 (20) study had bed lamps that were turned off at
Variables Mean ± SD‡ Mean ± SD Mean ± SD t p night after treatment, but some lights stayed
Age (years) 62⋅4 ± 11⋅8 59⋅9 ± 12⋅2 64⋅9 ± 10⋅9 1⋅67 0⋅92 on for patient observations. Continuous
APACHE II 15⋅4 ± 2⋅4 15⋅2 ± 2⋅6 15⋅7 ± 2⋅1 4⋅47 0⋅87 24-h treatment and nursing care, as well
Pain score 1⋅3 ± 0⋅8 1⋅3 ± 0⋅8 1⋅2 ± 0⋅8 0⋅67 0⋅96 as noises such as ringing phones, might
Heart rate (beats/min) 79⋅5 ± 7⋅6 79⋅7 ± 7⋅2 79⋅8 ± 7⋅3 35⋅9 0⋅88 be other reasons for poor sleep quality. The
Mean arterial pressure (mmHg) 97⋅8 ± 3⋅8 96⋅7 ± 4⋅4 99⋅5 ± 3⋅3 17⋅1 0⋅36 problem of poor sleep in the ICU seems to be
Respiratory rate (rate/min) 16⋅8 ± 2⋅3 17⋅3 ± 1⋅6 17⋅0 ± 3⋅0 25⋅1 0⋅97 a global issue (Bihari et al., 2012; Elliott et al.,
Subjective sleep scale§ 648 ± 25⋅3 637⋅7 ± 36⋅9 658⋅3 ± 13⋅9 25⋅7 0⋅04 2011; Little et al., 2012). A recent randomized
Objective sleep time (actigraphy)¶ 5⋅9 ± 0⋅7 5⋅7 ± 0⋅8 5⋅9 ± 1⋅8 1⋅6 0⋅10 controlled trial study examined the effects of
Objective sleep time (Nurses)∗∗ 4⋅0 ± 0⋅6 3⋅9 ± 0⋅5 3⋅9 ± 1⋅5 3⋅8 0⋅18‡ earplugs and eye masks and demonstrated
Anxiety†† 11⋅8 ± 1⋅5 12⋅1 ± 1⋅3 12⋅3 ± 1⋅8 11⋅4 0⋅79‡ their usefulness for promoting sleep (Hu
Depression‡‡ 9⋅0 ± 1⋅0 9⋅7 ± 2⋅1 9⋅5 ± 1⋅7 19⋅0 0⋅68‡ et al., 2015). However, while Demoule et al.
∗ Chi-square test. (2017) found a significant effect of earplugs
† p-value in Fisher’s exact test. and eye shades on sleep latency, not all
‡ SD, standard deviation. patients tolerated their use throughout the
§ Subjective sleep scale: Verran and Snyder-Halpern sleep scale, possible range (0–800). night. This highlights the need to decrease
¶ Objective sleep time: wrist actigraphy(h). the environmental sources of disturbance
∗∗ Objective sleep time: nursed observation (h). in the ICU. Furthermore, chronic diseases
†† Anxiety: Anxiety subscale of the Hospital Anxiety & Depression scale, possible range (0–21). might also lead to shorter sleep durations,
‡‡ Depression: Depression subscale of the Hospital Anxiety & Depression scale, possible range (0–21). which are consistent with the finding that
sleep quality worsens with an increase in
the number of chronic diseases in the elderly
GEE analysis indicated no significant dif- DISCUSSION (Wu et al., 2012).
ferences between the groups in improve- There is growing public concern about In this study, HR and mean arterial pres-
ment of anxiety scores from baseline to the the clinical practice to promote sleep in the sure were similar between the experimental
first or second day (p = 0⋅65; p = 0⋅50, respec- ICU. Information about the effects of back and control group at all time points. Con-
tively). In contrast, from baseline to the third massage on sleep effectiveness and sleep sistent with previous research (Bauer and
day, there was a significant difference in disruption is crucial for developing better Dracup, 1987), these results suggest that
improvement of anxiety between the groups and more appropriate non-pharmacological back massage has no obvious influence on
(p < 0⋅009; Figure S2E). This result indicates interventions. This study addressed a these parameters. However, there was a sig-
that back massage had significant effects on gap in the existing literature by exploring nificant improvement in RR because of back
anxiety by the third day of the interven- the effects of back massage in the ICU set- massage in patients on the third day of inter-
tion but no significant effect on depression ting, from an Asian perspective, in Chinese vention, which is again consistent with pre-
(Table 2). adults in Taiwan. In addition, this is only vious research (Chen et al., 2013). However,

4 © 2019 British Association of Critical Care Nurses

 Back massage and sleep quality

Table 2 Generalized estimating equation (GEE) for respiratory rate, sleep quality and anxiety (N = 60)

Subjective Objective sleep Observations of

RR sleep quality quality (Actigraphy, h) sleep (nurses, h) Anxiety
Variables B SE P B SE P B SE P B SE P B SE P

Intercept 16⋅81 0⋅68 <0⋅001*** 656⋅1 3⋅30 <0⋅001*** 6⋅00 0⋅10 <0⋅001*** 3⋅91 0⋅13 <0⋅001*** 12⋅45 0⋅34 <0⋅001***
Group (Exp. versus 0⋅86 0⋅95 0⋅37 −26⋅87 7⋅44 <0⋅001*** −0⋅41 0⋅25 0⋅01* 0⋅74 0⋅18 0⋅69 −0⋅29 0⋅48 0⋅64
control)
Time (day 1 versus pretest) 0⋅03 0⋅29 0⋅99 2⋅00 0⋅98 0⋅04* −0⋅12 0⋅08 0⋅16 −0⋅02 0⋅08 0⋅77 0⋅16 0⋅20 0⋅94
Time (day 2 versus pretest) −0⋅06 0⋅33 0⋅99 3⋅00 1⋅77 0⋅09 0⋅00 0⋅05 1⋅000 0⋅07 0⋅09 0⋅47 −0⋅20 0⋅23 0⋅40
Time (day 3 versus pretest) −0⋅19 0⋅29 0⋅50 4⋅67 2⋅04 0⋅02* 0⋅08 0⋅05 0⋅14 0⋅16 0⋅08 0⋅05 −0⋅41 0⋅21 0⋅045*
Group × time (day 1 −0⋅36 0⋅40 0⋅38 2⋅67 2⋅27 0⋅36 0⋅17 0⋅09 0⋅06 0⋅76 0⋅12 0⋅53 −0⋅13 0⋅29 0⋅65
versus pretest)
Group × time (day 2 −0⋅43 0⋅47 0⋅36 21⋅33 4⋅05 <0⋅001*** 0⋅30 0⋅12 0⋅04* 0⋅25 0⋅13 0⋅04* −0⋅23 0⋅33 0⋅50
versus pretest)
Group × time (day 3 −0⋅87 0⋅41 0⋅03* 31⋅33 5⋅48 <0⋅001*** 0⋅37 0⋅13 0⋅005** 0⋅26 0⋅12 0⋅03* −0⋅76 0⋅29 <0⋅009
versus pretest)

*p < 0⋅05; **p < 0⋅01; ***p < 0⋅001; RR = respiratory rate; B = estimated parameter; SE = standard error of coefficient; P = p-value.

RRs only declined slightly, with the mean RR
dropping by 0.9 breaths/min. Thus, while
there was a statistically significant change in
this study, it may not reflect clinical signifi-
cance.
It is important to note that, regardless
of subjective (VSH Sleep Scale scores)
and objective (wrist actigraphy and nurs-
ing observations) measurements, back
massage increased the patients’ TST and
improved their sleep quality; it particularly
decreased sleep latencies and increased
sleep effectiveness. In addition, significant
differences were seen in subjective sleep
quality and objective measures of sleep
duration between the groups on the second
day of intervention, suggesting that back
massage, even as a one-time intervention,
does have an effect on sleep quality and
TST. This result is consistent with several
studies that offered a 10–15-min back mas-
sage at night in the ICU (Shinde and Anjum,
2014) or cardiac ward for patients with
heart failure (Chen et al., 2013). All results
indicate that a massage releases muscle
tension, increases blood circulation and
initiates a relaxation response, resulting
in more restful sleep and thus improving
sleep quality and sleep duration. Impor-
tantly, back massage is a complementary
therapy and thus a non-pharmacological
sleep-promoting intervention that nurses
can apply in patient care. Such approaches
that emphasize the nurses’ independent
and autonomous role need to receive
more attention and be promoted in clinical
settings.
The current results show a difference in
measures of mean sleep duration between
nurse observations and wrist actigraphy.
Nonetheless, the changes observed in these
measures indicated an upward trend. The
nurses’ observations of sleep duration
were lower (0⋅65–0⋅68 times of the times
measured by wrist actigraphy), reflecting a
potential underestimation and suggesting a
need for further education of medical ICU
(MICU) nurses to improve their observa-
tions of sleep quality. Furthermore, future
studies are required to fully understand the
trajectory of sleep quality as observed by
nurses, including associated and potentially
confounding factors. Although actigraphy
is not a gold standard, like polysomnog-
raphy, it nevertheless is an objective and
relatively reliable instrument that can
be applied in future studies for investi-
gating sleep quality among critically ill
patients.
The results regarding anxiety in the
present study are consistent with those of
a previous study by Chen et al. (2013) that
observed a statistically significant difference
in improvements in anxiety between an
intervention group and a control group.
Gosselink et al. (2008) suggested that mas-
sages may improve muscle relaxation and
reduce pain, pressure and anxiety. A study
conducted by Chen et al. (2013) indicated
that back massage significantly reduces
anxiety in those with severe heart failure
and greater levels of anxiety.
However, not surprisingly, we found
no effect of back massage on depression.
Depression is a complicated disorder and is
related to a range of environmental factors,
as well as to heredity and early growth
experiences, personality characteristics, life
event stressors and a lack of social support.
These factors lead to alterations in endocrine
neurotransmitters in the brain, including
serotonin, dopamine and norepinephrine,
finally resulting in depression (Battle et al.,
2015; Jackson et al., 2014). Back massage
may not be effective enough to address
the influence of these numerous factors. In
addition, this study focused on ICU patients
who were generally older and had chronic
diseases, which might further influence
their depression symptoms.
LIMITATIONS
This study has several limitations. First,
we only examined ICU patients in a hospi-
tal setting, and the results might thus not
be generalizable to other patient popula-
tions. Furthermore, a ‘Hawthorne effect’
may have occurred in this study because
the experimental group received a back
massage, while the control group received
no comparable treatment. Finally, findings
show that sleep duration assessed by nurses
is often inconsistent with sleep duration
assessed by other objective measures. As
nurses only checked participants’ sleep
condition once every hour, and participants
might or might not have slept continu-
ously in between, the accuracy of these
assessments is limited. Hence, in future
studies, nurses should increase the duration

© 2019 British Association of Critical Care Nurses 5

Back massage and sleep quality
and/or frequency of observations for greater
validity.
IMPLICATIONS AND
RECOMMENDATIONS FOR
PRACTICE
Our findings may be useful for medical pro-
fessionals who manage ICU patients. Our
findings indicate that back massages are
safe and beneficial for critically ill patients.
Hence, we suggest that back massages
should be used to help improve sleep quality
in ICU patients. A back massage procedure
manual can be provided as a reference
for nurse educators and practicing nurses,
increasing their professional autonomy.
Furthermore, such interventions could be
provided to more diverse ICU populations,
although further investigation is needed.
WHAT IS KNOWN ABOUT THIS TOPIC
• Although sleep is important to promote recovery via effects on the immune system, alternative therapies such as back massages are not commonly used in the ICU to improve
patient sleep quality.
WHAT THIS PAPER ADDS
• A 10-minute back massage improves sleep quality, sleep duration, breathing and anxiety among ICU patients who have not undergone lung/abdominal surgery.
• The procedures and theoretical principles of massage interventions could be included in professional courses, particularly for critical care nursing.
REFERENCES
Altman MT, Knauert MP, Pisani MA. (2017).
Sleep disturbance after hospitalization
and critical illness: a systematic review.
Annals of the American Thoracic Society; 14:
1457–1468.
Battle C, James K, Temblett P. (2015). Depres-
sion following critical illness: analysis of
incidence and risk factors. Journal of Inten-
sive Care; 16: 105–108.
Bauer WC, Dracup KA. (1987). Physiologic
effects of back massage in patients with
acute myocardial infarction. Focus on Criti-
cal Care; 14: 42–46.
Bihari S, Doug McEvoy R, Matheson E, Kim S,
Woodman RJ, Bersten AD. (2012). Fac-
tors affecting sleep quality of patients in
intensive care unit. Journal of Clinical Sleep
Medicine; 8: 301–307. https://doi.org/10
.5664/jcsm.1920.
Boyko Y, Ørding H, Jennum P. (2012). Sleep
disturbances in critically ill patients in ICU:
how much do we know? Acta Anaesthesio-
logica Scandinavica; 56: 950–958. https://doi
.org/10.1111/j.1399-6576.2012.02672.x.
6 © 2019 British Association of Critical Care Nurses
CONCLUSION
This study bridges a gap in the existing
literature by investigating the effectiveness
of back massage in critical care settings.
Back massage is feasible for use in ICU
patients because it causes no adverse reac-
tions and improves sleep quality. Hence,
we suggest that the procedures and theo-
retical principles of massage interventions
be included in health care training, partic-
ularly for critical care nursing. However, its
cost-effectiveness remains to be evaluated.
ACKNOWLEDGEMENTS
The authors thank all the participants
in this study. This research received no
specific grant from any funding agency
in the public, commercial or not-for-profit
sectors.
Centers for Disease Control and Prevention.
(2017). Sleep and Sleep Disorders, Data
and Statistics, CDC, https://www.cdc
.gov/sleep/data_statistics.html (accessed
07/05/2018).
Chen WL, Liu GJ, Yeh SH, Chiang MC, Fu MY,
Hsieh YK. (2013). Effect of back massage
intervention on anxiety, comfort, and phys-
iologic responses in patients with conges-
tive heart failure. Journal of Alternative and
Complementary Medicine (New York, N.Y.); 19:
464–470.
Choi NG, DiNitto DM, Marti CN, Choi BY.
(2017). Too little sleep and too much
sleep among older adults: associations with
self-reported sleep medication use, sleep
quality and healthcare utilization. Geriatrics
& Gerontology International; 17: 545–553.
Delaney LJ, van Haren F, Lopez V. (2015).
Sleeping on a problem: the impact of sleep
disturbance on intensive care patients – a
clinical review. Annals of Intensive Care;
5: 3. https://doi.org/10.1186/s13613-015-
0043-2.
Demoule A, Carreira S, Lavault S, Pallanca O,
Morawiec E, Mayaux J, Arnulf I, Sim-
ilowski T. (2017). Impact of earplugs and
SUPPORTING INFORMATION
The following Supporting information is
available for this article:
Table S1. Back massage procedure.
Figure S1. Data collection process.
Figure S2. (A) Respiratory rates between
two groups. (B) Subjective sleep quality
is shown between two groups using the
Verran and Snyder-Halpern (VSH) Sleep
Scale. Lower scores represent worse qual-
ity of sleep. (C) Objective sleep duration
(hours) uses a wrist Actigraphy. (D) Objec-
tive sleep duration (hours) in two groups
is observed by nurses. (E) Anxiety scores
between two groups. A higher score implies
greater anxiety.
Additional supporting information may be
found online in the Supporting Information
section at the end of the article.
eye mask on sleep in critically ill patients: a
prospective randomized study. Critical Care
(London, England); 21: 284.
Elliott R, McKinley S, Cistulli P. (2011). The
quality and duration of sleep in the inten-
sive care setting: an integrative review.
International Journal of Nursing Studies;
48: 384–400. https://doi.org/10.1016/j
.ijnurstu.2010.11.006.
Elliott R, McKinley S, Cistulli P, Fien M. (2013).
Characterisation of sleep in intensive care
using 24-hour polysomnography: an obser-
vational study. Critical Care; 17: R46.
Fang CS, Wang RH. (2007). Sleep quality and its
associated factors among surgical intensive
care unit patients. Journal of Evidence-Based
Nursing; 3: 54–63.
Gaggioni G, Maquet P, Schmidt C, Dijk DJ,
Vandewalle G. (2014). Neuroimaging, cog-
nition, light and circadian rhythms. Fron-
tiers in Systems Neuroscience; 8: 126.
Gosselink R, Bott J, Johnson M, Dean E, Nava S,
Norrenberg M, Schönhofer B, Stiller K, van
de Leur H, Vincent JL. (2008). Physiother-
apy for adult patients with critical illness:

recommendations of the European Respira-
tory Society and European Society of Inten-
sive Care Medicine Task Force on physio-
therapy for critically ill patients. Intensive
Care Medicine; 34: 1188–1199. https://doi
.org/10.1007/s00134-008-1026-7.
Hu RF, Jiang XY, Chen J, Zeng Z, Chen XY,
Li Y, et al. (2015). Non-pharmacological
interventions for sleep promotion in the
intensive care unit. The Cochrane Database of
Systematic Reviews; 10: 1469–1493.
Jackson JC, Pandharipande PP, Girard TD,
Brummel NE, Thompson JL, Hughes CG,
Pun BT, Vasilevskis EE, Morandi A, Shin-
tani AK, Hopkins RO, Bernard GR, Dittus
RS, Ely EW, BRAIN-ICU (Bringing to
light the Risk Factors And Incidence of
Neuropsychological dysfunction in ICU
survivors) Study Investigators. (2014).
Depression, posttraumatic stress disor-
der, and functional disability in survivors
of critical illness in the BRAIN-ICU study:
a longitudinal cohort study. The Lancet.
Respiratory Medicine; 2: 369–379.
Lareau R, Benson L, Watcharotone K, Man-
guba G. (2008). Examining the feasibility
of implementing specific nursing inter-
ventions to promote sleep in hospital-
ized elderly patients. Geriatric Nursing; 29:
197–206.
LeGates TA, Fernandez DC, Hattar S. (2014).
Light as a central modulator of circadian
rhythms, sleep and affect. Nature Reviews
Neuroscience; 15: 443–454.
Liang KY, Zeger SL. (1986). Longitudinal data
analysis using generalized linear models.
Biometrika; 73: 13–22.
© 2019 British Association of Critical Care Nurses
Li L, Wu C, Gan Y, Qu X, Lu Z. (2016). Insomnia
and the risk of depression: a meta-analysis
of prospective cohort studies. BMC Psychia-
try; 16: 375–390.
Lin SL, Tsai SL. (2003). The reliability and
validity of Chinese version of Verran and
Snyder-Halpern Sleep Scale. VGH Nursing;
20: 105–106.
Little A, Ethier C, Ayas N, Thanachayanont T,
Jiang D, Mehta S. (2012). A patient survey of
sleep quality in the intensive care unit. Min-
erva Anestesiologica; 78: 406–414.
Polit DF, Beck CT. (2006). The content valid-
ity index: are you sure you know what’s
being reported? Critique and recommen-
dations. Research in Nursing & Health; 29:
489–497.
Pisani MA, Friese RS, Gehlbach BK, Schwab RJ,
Weinhouse GL, Jones SF. (2015). Sleep in
the Intensive Care Unit. American Journal of
Respiratory and Critical Care Medicine; 191(7):
731–738.
Richards K, Nagel C, Markie M, Elwell J,
Barone C. (2003). Use of complementary
and alternative therapies to promote sleep
in critically ill patients. Critical Care Nursing
Clinics of North America; 15: 329–340.
Salo P, Sivertsen B, Oksanen T, Sjosten N,
Pentti J, Virtanen M, Kivimaki M, Vahtera J.
(2012). Insomnia symptoms as a predic-
tor of incident treatment for depression:
prospective cohort study of 40,791 men and
women. Sleep Medicine; 13: 278–284.
Shinde MB, Anjum S. (2014). Effectiveness
of slow back massage on quality of
sleep among ICU patients. International
Back massage and sleep quality
Journal of Science and Research (IJSR); 3:
292–298.
Shiung TF, Chang YY, Chou SS, Chao YF.
(2009). An exploration on the related factors
of sleep quality in the intensive care unit.
VGH Nursing; 26: 274–281.
Snyder-Halpern R, Verran JA. (1987). Instru-
mentation to describe subjective sleep char-
acteristics in healthy subjects. Research in
Nursing & Health; 10: 155–163.
Turk DC, Dworkin RH, Trudeau JJ, Benson C,
Biondi DM, Katz NP, Kim M. (2015). Valida-
tion of the hospital anxiety and depression
scale in patients with acute low back pain.
The Journal of Pain; 16: 1012–1021.
Wu CY, Su TP, Fang CL, Chang MY. (2012).
Sleep quality among community-dwelling
elderly people and its demographic,
mental, and physical correlates. Jour-
nal of the Chinese Medical Association; 75:
75–80.
Xu T, Wick EC, Makary MA. (2016). Sleep
deprivation and starvation in hospitalised
patients: how medical care can harm
patients. BMJ Quality & Safety; 25: 311–314.
Yang Y, Ding R, Hu D, Zhang F, Sheng L. (2014).
Reliability and validity of a Chinese version
of the HADS for screening depression and
anxiety in psycho-cardiological outpatients.
Comprehensive Psychiatry; 55: 215–220.
Zigmond AS, Snaith RP. (1983). The hospital
anxiety and depression scale. Acta Psychi-
atrica Scandinavica; 67: 361–370. https://doi
.org/10.1111/j.1600-0447.1983.tb09716.x.
7