Presence and treatment of air hunger
in severely ill patients
Nathan M. Shumwaya,e,?, Ramey L. Wilsonb,e, Robin S. Howardc,
Joseph M. Parkerd,e, Arn H. Eliassond,e
aDepartment of Medicine, Brooke Army Medical Center, Ft. Sam Houston, TX 78258, USA
bDepartment of Medicine, Womack Army Medical Center, Ft. Bragg, NC 28310, USA
cDepartment of Clinical Investigation—Biometrics Service, Walter Reed Army Medical Center,
Washington, DC 20307-5001, USA
dDepartment of Medicine, Walter Reed Army Medical Center, Washington, DC 20307-5001, USA
eDepartment of Medicine, Uniformed Services University of the Health Sciences, Bethesda, MD 20814, USA
Received 21 May 2007; accepted 23 August 2007
Available online 1 November 2007
Air hunger at end-of-life poses challenges to providers who attempt to comfort while not
diminishing mental capacities.
We examined the presence, methods of assessment, and treatment of air hunger.
This observational study prospectively screened 198 consecutive medicine admissions for
increased risk of near-term death. These patients in turn were screened for dyspnea.
Patients screening positive were assessed on admission and the next day with the Visual
Analog Scale (VAS), modified Borg Scale, and the American Thoracic Society (ATS)
Shortness of Breath Scale. Additionally, resident physician opinions of patient dyspnea
level were assessed using the same tools. Treatments focused on alleviating air hunger
Thirty-nine percent of patients were at risk for near-term death and of these, 53% (95% CI:
41–65%) reported air hunger. All dyspnea scales improved to a statistically and clinically
significant degree (Borg p ¼ 0.007, VAS po0.0005, ATS p ¼ 0.008). There was statistically
significant agreement between Borg–VAS and between Borg–ATS with a trend toward
significance with ATS–VAS. Physician assessment of dyspnea showed poor agreement with
patients. A median of three treatments were received by patients but dyspnea
improvement did not correlate with the type, number, or specific combination of
ARTICLE IN PRESS
0954-6111/$-see front matter Published by Elsevier Ltd.
?Corresponding author. Hematology and Oncology Service, Department of Medicine, Brooke Army Medical Center, 3851 Roger Brooke Drive,
Ft. Sam Houston, TX 78234, USA. Tel.: +12109160504; fax: +12109163099.
E-mail address: firstname.lastname@example.org (N.M. Shumway).
Respiratory Medicine (2008) 102, 27–31
Dyspnea is common near end-of-life. Borg or VAS scales appear useful in assessing terminal
dyspnea and can be employed in assessing terminal air hunger. No individual treatment or
combination of treatments significantly improved patients’ dyspnea. However, air hunger
significantly improved with hospitalization.
Published by Elsevier Ltd.
Treating patients with air hunger at end-of-life, dyspnea
associated with the process of active dying, poses a special
challenge to providers who attempt to comfort without
hastening death or clouding the patient’s sensorium. In
2005, the American College of Chest Physicians released a
position statement regarding palliative and end-of-life care
for patients with cardiopulmonary diseases.1Although this
position statement serves as a framework for the manage-
ment of dyspnea, providers still struggle with how best to
assess and treat terminal dyspnea. One reason may be the
complex pathophysiology of dyspnea for which multiple
Choosing the right objective tools to assess air hunger can
be challenging as different tools can be more useful
depending on the clinical situation of the patient. Better
knowledge of the types of objective measurements available
may help providers care for their dyspneic patients. Mancini
and Body separate different assessment tools into three
categories: antecedent tools (physiologic or emotional
stressors that lead to the development of symptoms),
mediator tools (characteristics of individual or their
environment and how it affects response), and finally
reaction tools (outcomes that result once the stimulus has
been reached). Examples of antecedent tools are the British
Medical Research Council Scale (MRC), the American
Thoracic Society (ATS) questionnaire, and the Dyspnea
Interview Schedule. Mediator tools include the ATS, Chronic
Respiratory Questionnaire (CRQ), the Dyspnea Interview
Schedule, the Pulmonary Functional Status Scale (PFSS), and
the Therapy Impact Questionnaire (TIQ). The reaction tools
include the Visual Analog Scale (VAS), TIQ, modified Borg
Scale, the Dyspnea Interview Schedule, and the CRQ.3
Recently, two systematic reviews have been published
reviewing measurement scales to assess terminal air hunger.
Both concluded that while there is no single scale that can
adequately assess terminal dyspnea, some tools appear
more useful in the palliative setting. Dorman et al.4suggest
that the Numeric Rating Scale (NRS), modified Borg, Chronic
Respiratory Questionnaire Dyspnea subscale (CRQ-D), and
the Japanese Cancer Dyspnea scale (CDS) were best suited in
the palliative setting. Bausewein et al.5recommended
combining a one-dimensional scale such as the VAS with a
disease specific scale. Other options included use of a
multidimensional scale in conjunction with qualitative
assessment of psychosocial aspects of breathlessness. As
management of chronic illnesses becomes more sophisti-
cated and the prevalence of terminal dyspnea increases, the
ability to diagnose and treat air hunger will become
increasingly important as we seek to relieve suffering.
Air hunger is commonly associated with end-stage cancer.
The incidence of air hunger in terminal cancer patients in
the last 6 weeks of life is greater than 70%.6However, in
non-cancer patients dyspnea is also the most prevalent
respiratory symptom in the elderly near end-of-life.7
Despite being the most distressing symptom in dying
patients,8physicians have difficulty recognizing air hunger.
Such lack of awareness is hardly surprising in view of the
findings of the Study to Understand Prognoses and Prefer-
ences for Outcomes and Risks of Treatment (SUPPORT) which
showed that only 47% of physicians knew their patients’
preferences regarding something as fundamentally impor-
tant as CPR.9The difficulty in recognizing air hunger in
terminally ill patients has also been well described in the
intensive care unit setting.10Furthermore, physicians not
only fail to recognize the presence of dyspnea, but also do a
poor job estimating severity of this frightful symptom. While
data exists showing a lack of correlation with physician
assessments of dyspnea when compared to objective
measures such as brain naturetic peptide (BNP) for severity
of congestive heart failure, little data exist examining how
well physicians estimate severity of air hunger in terminally
Once patients with air hunger are identified, symptom
assessment tools are needed. Due to the complex patho-
physiology of this symptom more than one assessment tool
may be useful when treating patients with terminal
dyspnea.12These instruments should be simple to use for
both the patient and the provider. In addition, more than
one tool may be needed to assess for a more global picture
of the patient’s symptoms. For example, an antecedent tool
such as the ATS or MRC scale may help patients better
explain how anxiety or social stressors may contribute
to their air hunger. Mediator tools may help obtain a
more accurate assessment of how the patient’s environment
may contribute to severity of symptoms, and reaction tools
can help assess response to interventions for terminal
dyspnea. Use of these standardized tools can allow for
reproducible assessments of the severity of symptoms and
would greatly assist providers in the management of
terminal air hunger.
An earlier study at our institution identified that air
hunger in terminal patients was common and there was no
standardized approach to these patients. We also found that
no objective tools were used to quantify air hunger.13There
are few prospective studies of air hunger in terminal
patients and the results of our retrospective survey
stimulated the development of a prospective observational
study to assess how we care for patients with air hunger at
end-of-life. We specifically aimed to determine in a
prospective fashion the prevalence of air hunger in patients
with terminal conditions, to evaluate several measurement
tools in assessing air hunger, to determine the awareness of
the health care team as to their patient’s symptoms, and to
observe the impact of therapies on air hunger.
ARTICLE IN PRESS
N.M. Shumway et al.28
This prospective, observational study was conducted at
Walter Reed Army Medical Center, a military tertiary
care hospital. Beneficiaries eligible for care at WRAMC
include active duty service members, their dependents,
retired service members, and dependents of retirees. The
patient population is therefore comprised of men and
women of all ages, races, and a broad spectrum of cultural
For 2 months, consecutive admissions to the internal
medicine and hematology–oncology wards were evaluated
by reviewing the hospital’s database. As demographic
information was recorded, adult patients were evaluated
for severe illnesses which put them at risk for short-term
death. This was determined using a previously validated
screening tool (see Table 1).14Patients were excluded from
enrollment if they were less than 18 years old, unable to
give consent, and if they were admitted to the non-medical
services, bone marrow transplant ward, intensive care unit,
or coronary care unit.
Patients meeting enrollment criteria were asked for
informed consent to participate. All subjects enrolled
into the study voluntarily agreed to participate and
gave written informed consent. In this population of
severely ill patients, it was made clear to study subjects
that they could withdraw from any part of the research
study at any time. Once patient consent was obtained
a screening question was asked, ‘‘Are you having any
trouble with your breathing?’’ For those patients who
denied dyspnea, demographic data were obtained using
the computerized database and included co-morbid condi-
tions, age, gender, race, medications on admission, and
DNR/DNI status. Those patients who screened positive for
dyspnea were given questionnaires (day #1) which included
three different dyspnea scales: the VAS, the modified
Borg Scale, and the ATS Shortness of Breath Scale.
Additionally, the internal medicine resident caring for the
patient was asked to complete the same three dyspnea
scales based upon how they perceived their patient’s
symptoms upon admission to the hospital. On day #2,
patients were administered the same dyspnea question-
naires to assess for response to treatments administered for
air hunger. Treatments given were at the discretion of the
admitting physician and these treatments were recorded on
The Walter Reed Army Medical Center Clinical Investiga-
tion Committee and the Human Use Committee approved
Data analysis and statistics
Continuous and ordinal data are summarized using means or
medians with ranges, and selected proportions are pre-
sented with 95% confidence intervals (95% CI). Associations
among dyspnea scales and age are examined using Spear-
man’s correlation coefficient. Changes in dyspnea scales are
compared between groups (i.e. gender, race) using the
Wilcoxon rank sum test or Kruskal–Wallis analysis of
variance. Changes in dyspnea scales over time are compared
using the Wilcoxon signed ranks test.
There were 198 consecutive admissions to internal medicine
wards over a 2-month period. The mean age of patients was 62
years (range 19–97 years). Fifty-eight percent of patients were
Caucasian, 35% African-American, 4% Hispanic, and 4% Asian.
Admission diagnoses involved organ systems as follows: gastro-
enterology (21.7%), malignancy (17.1%), infection (17.1%),
cardiac (12.6%), pulmonary (9.5%), renal (6%), endocrine
(3%), neuro (2%), connective tissue (1.6%), and other (9%). Of
198 patients, 78 (39%) met criteria for being at risk for short-
term death. Of these 78 patients, 18 (23%) were unwilling to
participate in the study. Fifty-three percent (32/60) of enrolled
patients reported difficulty breathing (95% CI: 41–65%).
ARTICLE IN PRESS
Screening Tool11(used as screening tool to identify
patients at risk for short-term death).
Medical record Do-Not-Rescusitate (DNR)
1. CHF—NYHC III/IV
2. Valvular disease—NYHC
III/IV symptoms or
1. Dementia (end stage)
2. Progressive degenerative
disease (ALS, MS)
3. Severe structural deficit
1. COPD—cor pulmonale,
2. IPF or ILD
3. Pulmonary HTN that is
1. Collagen vascular
disease (end stage)
2. Vasculitis (PAN)
1. Cirrhosis (class III/IV,
Systemic diseases with poor
or no effective treatment
1. Chronic renal failure
(not candidate for
Extreme old age (485
1. AIDS—CD4 o50, poor
DNR order written
2. Hematologic disorders
(that no longer respond
Adapted from Archives of
Eliasson AH, Parker JM,
Shorr AF, Babb KA, Harris R,
Aaronson BA, Diemer M.
Impediments to writing do-
not-resuscitate orders. Arch
Air hunger in severely ill patients29
On admission median Borg, VAS, and ATS scores were 4/10,
4/10, and 3/4, respectively. All dyspnea scales improved
during hospitalization (Borg p ¼ 0.007, VAS po0.0005, ATS
p ¼ 0.008) and there was statistically significant agreement
among Borg–VAS and Borg–ATS with a trend toward
significance with ATS–VAS. Borg and VAS had excellent
agreement (Spearmans’ r ¼ 0.80, po0.0005, n ¼ 32). Borg–
ATS correlation was moderate (r ¼ 0.49, p ¼ 0.005), and
ATS–VAS association was weak (r ¼ 0.35, p ¼ 0.051).
Physicians’ initial assessment of their patient’s dyspnea
differed from patient scores by 3 points or more in Borg
11/32 (34%) and in VAS 16/32 (50%) (see Figure 1).
Treatments prescribed to treat patient’s air hunger
included inhaled b2-agonists and anticholinergics (69%),
oxygen (65%), antibiotics (38%), respiratory therapy (34%),
opiates (31%), tube thoracostomy (6%), thoracentesis (3%),
and others (18%).
On day #2, patients showed clinically and statistically
significant improvement in their dyspnea. Borg improved a
median of 1 point (range: ?3 to 8, p ¼ 0.007), VAS improved
1.3 points (range: ?4 to 7, po0.0005), and ATS improved
1.0 point (range: ?2 to 4, p ¼ 0.008). Improvement in Borg
did not correlate with age, race, gender, smoking history, or
specific treatments. Patients received a median of three
treatments and change in dyspnea score did not correlate
with the number of treatments received (see Figure 2).
The main finding of this study is a high prevalence (53%) of
air hunger in severely ill hospitalized patients. This is not
surprising given medical advances which allow patients to
live longer with their terminal conditions. Continued
advances in medicine, along with demographic trends,
will likely result in more patients presenting with dyspnea
late in the course of chronic terminal disease. The ability
of physicians to recognize and evaluate each patient’s
level of dyspnea and to choose effective treatments will
be important for optimal management of this complex
This study also showed that physicians do a poor job of
estimating the severity of dyspnea in their terminally ill
patients. Objective assessment tools can be very valuable
when used to evaluate air hunger. While there are many
assessment tools available, we feel that the modified Borg
Scale and the VAS provide reliable assessment tools that are
easy to administer to patients with dyspnea and to interpret
at the bedside. The agreement between these tools may
stem from the fact that the modified Borg and VAS are both
reaction tools. For terminally ill patients, reaction tools
provide an effective and rapid way to evaluate changes in
sensed work of breathing. As terminal dyspneic patients are
often full of fear and anxiety, these tools can be ideal and
comforting to patients, as they allow them to quickly relay
to their provider the severity of symptoms and response to
therapies given. These findings are consistent with the
recent systematic reviews by Bausewein et al. and Dorman
et al. regarding measurement tools for assessment of air
hunger in the palliative setting. In our study, the ATS tool
was not as useful. The ATS tool is both an antecedent and
mediator tool. Perhaps the poor agreement of ATS with the
other tools was due to the fact that it was compared to two
reaction tools. We also found that ATS did not have the same
ease of use in our severely ill patient population.
The lack of correlation between improvement in dyspnea
and the number of treatments given suggests that treat-
ments are best when individualized and tailored to each
patient’s specific medical conditions. Alternatively, the
improvement in dyspnea may be secondary to factors not
measured in this study. Perhaps bed rest alone led to
improvement in dyspnea, or perhaps sources of stress from
outside of the hospital were diminished.
A limitation of this study is that it was performed at a
tertiary care institution. Prevalence of air hunger might be
different in a community-based, non-referral center. An-
other limitation is the exclusion of sicker patients who were
unable to give consent. Inclusion of these patients may have
demonstrated greater impact of therapy. Another poten-
tially confounding factor stemmed from an ethical concern.
If patients stated that they had problems with their
breathing, this information was communicated to the
medical team caring for the patient. This communication
may have influenced therapeutic decisions by the team.
The dyspnea tools are not validated for surrogate ratings.
However, the failure of physicians to estimate dyspnea
ARTICLE IN PRESS
Patient VAS Score
Physician VAS Score
Line of Agreement
compared to patient VAS score.
Physician assessment of air hunger using VAS
Number of Treatments
Change in Borg Score
recorded treatments received.
Change in Borg score compared to number of
N.M. Shumway et al.30
reliably suggests that objective tools should be used to Download full-text
measure the symptom and to guide therapy. Although reliable
objective tools exist, these instruments have not been part of
clinical practice. We need to better educate health care
personnel as to the prevalence of air hunger and make the
objective tools both readily available and easy to use.
In summary, this study demonstrates that air hunger is
prevalent in patients with terminal disease who are
admitted to the hospital. Patient care will likely benefit
from the use of objective measurement tools to assess the
level of dyspnea. The modified Borg Scale and VAS are useful
tools for this evaluation and we recommend their use.
Patients’symptoms improve with hospitalization but there is
no correlation between degree of improvement and the
number or type of treatments implemented. Improvement
may be due to multiple factors not measured in this study.
Given its striking prevalence, air hunger and effective
therapy for this frightening symptom deserve further study.
Conflict of interest statement
None of the authors have a conflict of interest to declare in
relation to this work.
1. Selecky PA, Eliasson AH, Hall RI, Schneider RF, Varkey B.
Palliative and end-of-life care for patients with cardiopulmon-
ary diseases: American College of Chest Physicians Position
Statement. Chest 2005;128:3599–610.
2. Dyspnea. Mechanisms, assessment, and management: a con-
sensus statement. American Thoracic Society. Am J Respir Crit
Care Med 1999;159:321–40.
3. Mancini I, Body JJ. Assessment of dyspnea in advanced cancer
patients. Support Care Cancer 1999;7:229–32.
4. Dorman S, Byrne A, Edwards A. Which measurement scales
should we use to measure breathlessness in palliative care? A
systematic review. Palliat Med 2007 Mar 15; [Epub ahead of
5. Bausewein C, Farquhar M, Booth S, Gysels M, Higginson IJ.
Measurement of breathlessness in advanced disease: a systema-
tic review. Respir Med 2007;101(3):399–410.
6. Reuben DB, Mor V. Dyspnea in terminally ill cancer patients.
7. Jacobs LG. Managing respiratory symptoms at the end of life.
Clin Geriatr Med 2003;19:225–39.
8. Steinmetz D, Walsh M, Galbel LL, Williams PT. Family
physicians’ involvement with dying patients and their families.
Attitudes, difficulties, and strategies. Arch Fam Med 1993;
9. A controlled trial to improve care for seriously ill hospitalized
patients: The Study to Understand Prognoses and Preferences
for Outcomes and Risks of Treatments (SUPPORT). The SUPPORT
Principal Investigators. JAMA 1995;274:1591–8.
10. Cambell ML. Terminal dyspnea and respiratory distress. Crit
Care Clin 2004;20(3):403–17 viii–ix.
11. McCullough PA, Nowak RM, McCord J, et al. B-type natriuretic
peptide and clinical judgment in emergency diagnosis of heart
failure: analysis from Breathing Not Properly (BNP) Multi-
national Study. Circulation 2002;106(4):416–22.
12. Van der Molen B. Dyspnoea: a study of measurement instru-
ments for the assessment of dyspnoea and their application for
patients with advanced cancer. J Adv Nurs 1995;22:948–56.
13. Shumway NM, Kelly WF, Moquin ME, Eliasson AH. Treatment of
air hunger at end of life [abstract]. Am J Respir Crit Care Med
14. Eliasson AH, Parker JM, Shorr AF, Babb KA, Harris R, Aaronson
BA, et al. Impediments to writing do-not-resuscitate orders.
Arch Intern Med 1999;159:2213–8.
ARTICLE IN PRESS
Air hunger in severely ill patients31