Professional Documents
Culture Documents
Original Article
Abstract
Context. Dyspnea is a common distressing symptom among patients with advanced cancer.
Objective. The objective of this study was to determine the effect of fan therapy on dyspnea in patients with terminally ill cancer.
Methods. This parallel-arm, randomized controlled trial included 40 patients with advanced cancer from a palliative care
unit at the National Cancer Center Hospital in Japan. All patients experienced dyspnea at rest with a score of at least three
points on a subjective 0- to 10-point Numerical Rating Scale (NRS), showed peripheral oxygen saturation levels of $90%, had
an Eastern Cooperative Oncology Group grade of 3 or 4, and were aged 20 years or more. In one group, a fan was directed to
blow air on the patient’s face for five minutes. This group was compared to a control group wherein air was blown to the
patient’s legs. Patients were randomly assigned to each group. The main outcome measure was the difference in dyspnea NRS
scores between fan-to-face and fan-to-legs groups.
Results. No significant differences were seen in baseline dyspnea NRS between groups (mean score, 5.3 vs. 5.1, P ¼ 0.665).
Mean dyspnea changed by 1.35 points (95% CI, 1.86 to 0.84) in patients assigned to receive fan-to-face and by 0.1
points (0.53 to 0.33) in patients assigned to receive fan-to-legs (P < 0.001). The proportion of patients with a one-point
reduction in dyspnea NRS was significantly higher in the fan-to-face arm than in the fan-to-legs arm (80% [n ¼ 16] vs. 25%
[n ¼ 5], P ¼ 0.001).
Conclusion. Fan-to-face is effective in alleviating dyspnea in patients with terminally ill cancer. J Pain Symptom Manage
2018;56:493e500. Ó 2018 American Academy of Hospice and Palliative Medicine. Published by Elsevier Inc. All rights reserved.
Key Words
Dyspnea, neoplasms, palliative care, randomized controlled trial, nursing
Address correspondence to: Jun Kako, MHSc, RN, OCNS, Divi- Minami-ku, Hiroshima, 734-8533, Japan. E-mail: jkako-tky@
sion of Nursing Science, Graduate School of Biomedical umin.ac.jp
and Health Sciences, Hiroshima University, 1-2-3, Kasumi, Accepted for publication: July 2, 2018.
Ó 2018 American Academy of Hospice and Palliative Medicine. 0885-3924/$ - see front matter
Published by Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jpainsymman.2018.07.001
494 Kako et al. Vol. 56 No. 4 October 2018
Introduction Methods
Dyspnea, a common and distressing symptom Study Design
among patients with advanced cancer, is defined as We conducted a parallel-arm RCT (Japanese Clin-
an unpleasant or uncomfortable sensation during ical Trials Register UMIN000023345; https://upload.
breathing.1,2 This symptom has negative physical, umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno¼R
emotional, and psychosocial effects. The prevalence 000026902) and recruited patients between
of dyspnea increases as patients approach death,3,4 September 28, 2016, and August 25, 2017. We adopted
and its alleviation is, therefore, critically important a parallel control design because in our pilot study,
to improve the quality of life in such patients. Manage- the washout period for fan therapy that may affect
ment of dyspnea includes treatment of the underlying the study outcome needed to be more than one
causes with various combinations of pharmacological hour, which indicated that a crossover design was inap-
therapy and nonpharmacological approaches.5e7 propriate.17,20 Fig. 1 shows the CONSORT flowchart
Fan therapy, often used to palliate dyspnea, uses a for patient selection.
fan to blow air in the direction of the patient’s face.
The mechanism by which fan therapy provides relief Participants and Settings
from symptoms of dyspnea is not yet clear; however, Participants were recruited from a palliative care
direct stimulation of the face, nasal mucosa, or phar- unit of the National Cancer Center Hospital East, Chi-
ynx, as well as changes in facial temperature by cool- ba, Japan. Patients were required to meet the
ing due to the airflow, may affect the ventilation following inclusion criteria: 1) metastatic or locally
patterns.8e10 advanced cancer, 2) not undergoing current or
Although fan therapy is recommended by various further anticancer treatment, 3) dyspnea while sitting
clinical guidelines,11e13 there is limited empirical ev- or lying at rest with a score of at least three points on a
idence to support its effectiveness.14,15 A randomized 0- to 10-point Numerical Rating Scale (NRS) (0 ¼ no
trial conducted by Galbraith et al. reported that fan breathlessness; 10 ¼ worst possible breathlessness); 4)
therapy was effective in reducing dyspnea, but the peripheral oxygen saturation levels $90%; 5) Eastern
subjects studied were patients with different primary Cooperative Oncology Group grade of 3 or 4; 6)
advanced diseases, and included 11 patients with pri- aged $ 20 years; and 7) no cognitive impairment
mary or secondary lung cancer.16 Galbraith et al. con- and able to communicate in Japanese. Owing to the
ducted a second single-arm study targeting a similar lack of an established definition, we defined our target
population (n ¼ 31) and reported that half of the population (patients with terminally ill cancer) using
participants showed reduced intensity of dyspnea the criteria of disease (metastatic or locally advanced),
with fan therapy.17 Other randomized controlled tri- treatment (no anticancer treatment), and perfor-
als targeting patients with cancer have involved small mance status. The exclusion criteria were fever
sample sizes (n ¼ 21) with varying performance sta- >38 C in the preceding 24 hours, a hemoglobin level
tus.18 Wong et al.19 in a randomized controlled trial #6 g/dL, and diseases or treatments affecting the tri-
(RCT) of patients with terminally ill cancer reported geminal nerve.
on the effectiveness of fan therapy. However, this
study had neither an adequate sample power analysis Interventions and Procedures
nor a clear definition of the target population. Allevi- Fan therapy constituted of directing a fan to blow
ating dyspnea in patients with terminally ill cancer is air for five minutes across the region innervated by
a crucial issue; therefore, it is important to evaluate the second/third trigeminal nerve branches. The fan
the effectiveness of fan therapy in these patients. (model PJ-B3CLL [SHARP, Sakai-ku, Sakai, Japan];
However, adequately powered RCTs have not been five blades; size, 37 35.6 84 cm) was directed to-
conducted to examine the efficacy of fan therapy ward one side of the face. The rationale for using
for dyspnea in such patients. Given the minimal po- five minutes of directed airflow was based on previous
tential side effects, low cost, practical convenience, findings that this protocol achieved symptom pallia-
and immediate responses, a well-designed clinical tion.16,17,20 As in previous studies,16 the distance, loca-
trial evaluating the effectiveness of fan therapy will tion, side of the face, strength, and swing of the fan
be of great value. were determined as per the patient’s prefer-
The primary aim of this study was, therefore, to eval- ences.16,18,20 A standing fan placed on the floor was
uate the effectiveness of fan therapy for dyspnea in pa- applied at the lowest speed initially and was gradually
tients with terminally ill cancer. In addition, we aimed adjusted to increase the speed and strength of the fan
to investigate the changes in patients’ facial surface breeze. In the control arm, airflow was directed onto
temperature and physiological parameters, after fan the legs with the patient’s skin exposed for
therapy. five minutes using the same model of fan as that
Vol. 56 No. 4 October 2018 Fan Therapy for Dyspnea in Patients With Cancer 495
used in the intervention arm. We adopted the fan-to- as a minimal clinically important difference was defined
legs method as the control treatment based on the as a one-point reduction in the NRS score.24e27
study by Galbraith et al.16 The secondary outcomes included changes in pa-
We applied a washout time based on the NCCN tients’ facial surface temperature, measured using a
guidelines before initiating therapy so as to avoid FLIRÒ TG165 IR thermometer (FLIR Systems Inc.,
the effects of prior opioid treatment on the study Wilsonville, OR) and other physiological parameters
outcome.21 such as the respiratory rate, peripheral oxygen satura-
tion level, and pulse rate. Ambient room temperature
Outcomes and humidity were also measured. These data were re-
The primary outcome was change in dyspnea NRS. corded at baseline and immediately after the treat-
The severity of dyspnea was recorded at baseline and ment. We did not conduct the arterial blood gas and
immediately after treatment, using a Japanese version respiratory function test due to the patient burden
of the Edmonton Symptom Assessment Systemerevised this test imposes. The investigator monitored adverse
(ESAS-r).22 The ESAS-r measures nine symptoms that events by directly questioning the patient if adverse
include pain, tiredness, drowsiness, nausea, lack of appe- events had been experienced. Such information was
tite, depression, anxiety, dyspnea, and well-being.23 The reported to the concerned palliative care specialists.
ESAS-r rates these symptoms on an 11-point Likert scale Baseline characteristics such as primary tumor sites,
from 0 (no symptoms) to 10 (most severe symptoms). For comorbidities, underlying etiologies of dyspnea,
sensitivity analyses, we conducted an exploratory Eastern Cooperative Oncology Group grade, Karnof-
responder analysis based on the proportion of patients sky Performance Status score, Palliative Prognostic
in both the study groups who experienced a $1 point/ Score, Palliative Prognostic Index, and prescriptions
$2 point reduction, and $10%/25% reduction in the for opioids, steroids, and oxygen were obtained from
dyspnea NRS. The clinical benefit for dyspnea measured patient medical records.28,29 Comorbidities and
496 Kako et al. Vol. 56 No. 4 October 2018
underlying etiologies were determined based on clin- Patient Characteristics and Baseline Symptoms
ical judgments made by the palliative care specialists Patient characteristics were generally similar be-
primarily responsible for the patient’s care. tween the intervention and control arms (Table 1).
The average age was 69 years, and 22 of 40 (55%) pa-
Sample Size tients were men. Approximately two-thirds of the par-
We estimated that >16 patients per group would ticipants had primary or secondary lung cancer. The
allow for detection of a mean difference of 1.0 (SD average hemoglobin level was 10.9 g/dL. Types of
1), with a P-value of <0.05 at a power of 80%, using opioids used included morphine, oxycodone, and
the dyspnea NRS. The SD value was decided based fentanyl, with a mean daily oral morphine dose of
on the results of our pilot study.20 To allow for a con- 30.3 mg/d (SD 31.9, range 0e119 mg/day). About
servative estimated attrition of 20%, we planned to re- 80% of the participants had a Karnofsky Perfor-
cruit at least 40 patients. mance Status score of 40 or less indicating a disability
for self-care. While 52.5% (n ¼ 21) had a Palliative
Randomization Prognostic Score of 9 or more, 70% (n ¼ 28) had a
Participants who met all the eligibility criteria and Palliative Prognostic Index score of 6.5 or more.
provided written informed consent were randomly as- Half of the participants were on supplemental oxy-
signed in a 1:1 ratio to the intervention (fan therapy: gen at the time of enrollment, with a median of
fan-to-face) or to the control group (fan-to-legs). Par- 2 L/minute and an average oxygen saturation of
ticipants were stratified based on baseline dyspnea 96% (SD 2).
NRS levels (#7) and randomized using a software Table 2 shows the intensity of the baseline symptoms
application available (via https://epocdatabase.epoc- measured by the ESAS-r. No significant differences
ncc.net/) through the clinical support system at the were seen in the baseline dyspnea NRS between the
National Cancer Center. intervention and the control groups (mean score,
5.3 vs. 5.1, P ¼ 0.665).
Statistical Analysis
All randomized patients who completed a baseline
Changes in Dyspnea and Other Symptom Intensities
Table 3 summarizes the changes in dyspnea inten-
assessment were included in the primary analysis.
sity. The mean scores changed by 1.35 points (95%
We analyzed the descriptive statistics for patient char-
CI, 1.86 to 0.84) in the fan-to-face group versus
acteristics. Continuous variables were compared using
by 0.1 points (CI, 0.53 to 0.33) in the fan-to-legs
the Student’s t-test, and categorical variables were
group (P < 0.001). The proportion of patients with
compared using the c2 test or the Fisher’s exact
$1-point and $2-point reductions in dyspnea NRS
test, as appropriate. For the primary outcome, the dif-
scores was significantly higher in the fan-to-face group
ference in dyspnea NRS between fan-to-face and
than in the fan-to-legs group: 80% versus 25%,
fan-to-legs was assessed using the Student’s t-test. For
P ¼ 0.001; 35% versus 5.0%, P ¼ 0.043 (Table 3). Simi-
secondary analyses, we defined a P-value <0.05 as sta-
larly, the proportion of patients with $10% reduction
tistically significant, due to expletory nature of these
was significantly higher in the fan-to-face group than
outcomes.
in the fan-to-legs group: 80% versus 25%, P ¼ 0.001
Statistical analyses were performed using the EZR
(Table 3).
statistical software (Saitama Medical Center, Jichi Med-
The change in the drowsiness score was significantly
ical University, Saitama, Japan) and R (The R Founda-
higher in the fan-to-face group than in the fan-to-legs
tion for Statistical Computing, Vienna, Austria).30
group (absolute difference: þ0.40) (Table 4). The
other symptoms were not different between the
groups (Table 4).
Results No adverse effects were reported to the palliative
Between September 28, 2016, and August 25, 2017, care specialists primarily responsible for patient care.
we screened 429 eligible patients. Of 429 enrolled pa-
tients, 389 declined to participate in the RCT. Except Changes in Facial Surface Temperature and
for one patient who was not interested, the rest of the Physiological Parameters
patients did not meet inclusion criteria. A total of 40 Facial surface temperatures at baseline were not
patients were included and were randomly assigned significantly different between the two groups (fan-
to the two groups: 20 patients to the fan-to-face and to-face group vs. fan-to-legs group: 33.2 [SD 1.5]
20 to the fan-to-legs groups (Fig. 1). All participants vs. 33.2 [SD 1.8], P ¼ 0.95). However, the fan-to-
completed the study. face group showed a drop in temperature (by
Vol. 56 No. 4 October 2018 Fan Therapy for Dyspnea in Patients With Cancer 497
Table 1
Patient Characteristics (n ¼ 40)
Variable Fan-to-Face (n ¼ 20) Fan-to-Legs (n ¼ 20) P-value
1.43 C) after the intervention (95% CI, 2.30 to Changes in physiological parameters including
0.56), which was significantly different from the pulse rate, respiratory rates, and SpO2 showed no
change in the fan-to-legs group (0.01-point significant differences between the two groups
decrease; 95% CI, 0.36 to 0.34; P ¼ 0.003). (Table 5).
Table 2
Baseline Symptom Intensity Measured by the Edmonton Symptom Assessment SystemeRevised
Symptom Fan-to-Face Group (n ¼ 20) Fan-to-Legs Group (n ¼ 20) P-value
Table 3
Changes in Dyspnea Intensity
Dyspnea Score Fan-to-Face Group (n ¼ 20) Fan-to-Legs Group (n ¼ 20) P-value
Absolute change (95% CI) 1.35 (1.86 to 0.84) 0.10 (0.53 to 0.33) <0.001
One-point reduction, n (%) 16 (80.0) 5 (25.0) 0.001
Two-point reduction or more, n (%) 7 (35.0) 1 (5.0) 0.043
Relative change (%) 27.7% 1.7% 0.002
$10% reduction, n (%) 16 (80.0) 5 (25.0) 0.001
$25% reduction, n (%) 8 (40.0) 2 (10.0) 0.065
Table 4
Changes in the Intensity of Symptoms Other Than Dyspnea
Symptom Fan-to-Face Group (n ¼ 20) Fan-to-Legs Group (n ¼ 20) P-value
Table 5
Changes in Physiological Parameters
Fan-to-Face Group (n ¼ 20) Fan-to-Legs Group (n ¼ 20)
Pulse rate 93.9 (17.4) 88.0 (25.3) 0.114 94.3 (17.3) 92.4 (17.3) 0.125
Respiratory rate 17.8 (4.8) 17.6 (4.6) 0.522 14.6 (4.2) 14.7 (4.2) 0.716
SpO2 95.7 (2.0) 95.7 (2.1) 0.858 96.7 (2.0) 96.9 (1.8) 0.408
SpO2 ¼ peripheral oxygen saturation levels.
Means (SDs) are presented.