ttp://www.bsava.

com PAPER

Rapid assessment with physical

examination in dyspnoeic cats:
the RAPID CAT study
D. Dickson*,1, C. J. L. Little†, J. Harris‡ and M. Rishniw§

*HeartVets, 7 Kingshill, Porthcawl, CF36 5LD, UK

†
Barton Veterinary Hospital, 34 New Dover Road, Canterbury, CT1 3AT, UK
‡
HeartVets, Exeter, EX4 2HP, UK
§
Veterinary Information Network, 777 West Covell Boulevard, Davis, California 95616, USA
1
Corresponding author email: dave@heartvets.co.uk

OBJECTIVES: To determine the underlying disease prevalence in acute feline dyspnoea and to examine
whether historical and clinical examination findings can differentiate between acute cardiac and non-
cardiac dyspnoea in cats in primary practice.
MATERIALS AND METHODS: We prospectively enrolled cats presenting with dyspnoea for the first time to
primary practice between June 1, 2011 and October 31, 2016. We collected signalment, historical
and clinical data at presentation using a standard form. Cases were investigated by primary clinicians,
and the final diagnosis was confirmed by the authors. Records lacking critical data were excluded.
Relationships between historical or clinical variables and dyspnoea aetiology were examined. Diagnostic
test performance analyses were used to find optimal cut-off values for select historical or clinical
variables that could differentiate cardiac and non-cardiac dyspnoea.
RESULTS: Participants included 108 cats. A definitive diagnosis was reached in 92 cases; 60 were
cardiac (65%), 15 respiratory (16%), 10 neoplastic (11%) and 7 traumatic (8%). Of cats with cardiac
dyspnoea, 25% had a history of cough. A gallop sound, rectal temperature less than 37·5°C, heart
rate of greater than 200 bpm and respiratory rate greater than 80 per minute were all useful to predict
cardiac-associated dyspnoea. A triage algorithm using these findings in combination was designed to
optimise the rapid diagnosis of probable cardiac dyspnoea.
CLINICAL SIGNIFICANCE: Dyspnoeic cats presenting in practice with hypothermia, tachycardia, gallop sounds
or profound tachypnoea are likely to have a cardiac cause underlying their dyspnoea. Although diagnosis
requires confirmation, clinicians may be able to stratify risk and prioritise further investigation based
on these findings.

Journal of Small Animal Practice (2017)

DOI: 10·1111/jsap.12732
Accepted: 20 July 2017

INTRODUCTION observant owners, and thus, subtle signs of a progressive disease

Dyspnoea, defined as difficult or laboured breathing, is a com-
mon presentation in veterinary practice and can be associated
with life-threatening diseases. Cats can provide a particular
challenge: they are adept at concealing clinical signs even from
may be overlooked. Cats commonly present in extremis, and in
such situations, rapid and accurate diagnosis of the underlying
cause of the dyspnoea is imperative for appropriate patient man-
agement; delayed or inappropriate action may compromise the
cat or even prove fatal. Consequently, easily and rapidly obtained

Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association 1
 D. Dickson et al.
clinical information that has an established value to differentiate
likely causes of dyspnoea is valuable to clinicians.
Most small animal clinicians in primary care practice are familiar
with the presentation of cats with acute-onset dyspnoea, without a
previous history of respiratory difficulties. Experience has shown
that these cats are often physiologically very fragile (Mandell 2004,
Kellett-Gregory & King 2010, Ferasin & DeFrancesco 2015).
Some cats are so vulnerable that otherwise routine handling and
investigative procedures can precipitate further decompensation or
even death. Many different diseases can cause dyspnoea or respira-
tory distress in cats, but there is little published evidence examining
the prevalence of these various diseases in the general cat popula-
tion. Indeed, evidence appears limited to one recent peer-reviewed
retrospective study of dyspnoea in cats presenting to a referral prac-
tice in the UK (Swift et al. 2009). However, it is very likely that
those cats who survive long enough to be referred are not typical
of dyspnoeic cats seen in primary care practice. We are aware of
only one study examining cardiovascular disease in cats in primary
care practice, which included dyspnoeic patients (Smith & Dukes-
McEwan 2012). However, that study contained both primary and
referral practice cases, and fewer than half of the cats were dyspnoeic
on presentation. The authors concluded that certain clinical signs
were useful in identifying cardiac dyspnoea, but the mix of primary
and referral cases and mix of presentations in that study might limit
the generalisability of the findings if they are applied to managing
acute dyspnoea in primary care practice.
Many primary care practices lack advanced diagnostic facili-
ties, and moreover, dyspnoeic cats generally benefit from minimal
handling and disturbance. Therefore, findings based on history
and limited physical examination that can provide primary care
clinicians with sufficient information to recognise the most likely
underlying cause of dyspnoea and to implement initial, poten-
tially life-saving treatments would be of great value. Indeed, Swift
et al. (2009) suggested that basic historical and clinical examina-
tion findings might be useful to predict the cause of dyspnoea in
cats, although whether such findings can be extrapolated to cases
in primary care practice is unknown.
Therefore, we wanted to establish the relative frequencies of the
causes of dyspnoea observed in primary practice and determine
the extent to which certain historical and clinical findings were
specifically associated with particular causes of acute dyspnoea in
cats. We then investigated whether a combination of such find-
ings would improve diagnostic utility, specifically to discriminate
between cardiac and non-cardiac causes of dyspnoea.
MATERIALS AND METHODS
Recruitment and data collection
Cats that were presented with acute dyspnoea to any of nine
primary care practices in the UK were prospectively recruited
between June 2011 and October 2016. Cats were examined
by primary clinicians (n=27); case records and diagnoses were
reviewed by the supervising clinician (authors, who are either
board-certified cardiologists or residents in training, under super-
vision of a board-certified cardiologist). To be included, cats must
2 Journal of Small Animal Practice
have been presented for the first time with dyspnoea (defined as
difficult or laboured breathing) and could not be on treatment
for known thoracic disease. Cats were included in the study if
they were presented to a primary care practice veterinarian with
dyspnoea being the main complaint or if, on initial assessment,
the clinician immediately noted dyspnoea. Owners provided
informed consent as soon as was practical and before data were
recorded in the study database. Owner and patient details were
not shared outside the study group, and primary clinicians did
not have access to other clinicians’ data.
All clinicians were given a study protocol to standardise the
data collected, using a predesigned data-capture form, with infor-
mation recorded at the time of initial presentation. Clinicians
were asked to include every dyspnoeic cat, except if clients
refused consent. Primary clinicians recorded clinical findings
upon patient presentation or during the first few minutes of
patient stabilisation, as soon as was deemed safe, before medi-
cal management or further diagnostic testing. Oxygen therapy
was provided as soon as necessary, before collection of clinical
data if needed. Patient age, gender and breed and the follow-
ing historical data were recorded: duration of dyspnoea (hours),
history of cough (yes/no), wheeze (yes/no), weight loss (yes/no),
recent history of trauma (yes/no) and changes in appetite or thirst
(yes/no). Any other history the clinician deemed pertinent to the
dyspnoea was also noted. Clinical examination records included:
rectal temperature (T; °C), heart rate/pulse rate (P; beats/minute,
recorded over at least 30 seconds), respiratory rate (R; breaths/
minute), respiratory effort (normal/increased), mouth breathing
(yes/no), abdominal effort (yes/no), mucous membrane colour
(red/pink/pale/white/blue), cardiac auscultation (normal/mur-
mur/gallop/arrhythmia) and pulmonary auscultation (normal/
abnormal). Any other findings the clinician deemed pertinent to
the dyspnoea were also recorded. Grades or categories of cardiac
murmurs were not requested.
Classification of cases
The diagnosis (when obtained) was recorded. Diagnoses were
grouped as cardiac [i.e. congestive heart failure (CHF)], respira-
tory, neoplastic, traumatic or miscellaneous/undiagnosed, the last for
diagnoses that did not fit another category or if a definitive diag-
nosis was not obtained. All diagnoses made by primary clinicians
were reviewed and confirmed by the supervising clinician follow-
ing patient stabilisation or discharge. If the clinicians were unable
to make a diagnosis, no diagnosis was recorded unless informa-
tion gained from case follow-up allowed a definitive diagnosis
to be made. This was a prospective observational study, so each
case was investigated and managed as deemed appropriate by the
primary and supervising clinicians; tests were chosen based on
relevance to the case, but test selection was influenced by client
preference and financial resources as well as the experience of the
primary clinician.
Cats undergoing echocardiography were considered by the
authors to be in CHF (cardiac) if all the following criteria were met:
• Echocardiography indicated increased left atrial size [short-axis
left atrium to aortic (LA:Ao) ratio>1·5:1 or left atrial long-
• © 2017 British Small Animal Veterinary Association
 RAPID CAT - prospective study of feline dyspnoea
axis dimension>16·5 mm; Abbott & MacLean 2006, Smith &
Dukes-McEwan 2012] and evidence of primary cardiac disease.
• Echocardiography or radiography indicated pleural effusion or
radiographic changes consistent with CHF.
• There was a good response to therapy directed at managing acute
CHF (diuresis, oxygen supplementation or thoracocentesis).
In cases in which echocardiography was not performed, a diag-
nosis of CHF was only made if the following criteria were met:
• Compatible radiographic findings indicative of heart failure,
such as marked cardiomegaly, plus evidence of pulmonary
oedema or a thoracic effusion with a total protein content of
greater than 20 and less than 50 g/L.
• The cat showed a rapid and substantial improvement following
therapy directed at managing acute CHF, or there was con-
vincing post mortem evidence of heart disease and pulmonary
oedema or pleural effusion.
Respiratory dyspnoea was diagnosed based on radiographic,
bronchoscopic, cytological or histological evidence of significant
respiratory tract disease (excluding neoplasia). This included
bronchial and interstitial pulmonary disease as well as pyothorax,
airway obstruction and foreign bodies.
Neoplastic dyspnoea was diagnosed based on radiographic,
bronchoscopic, cytological or histological evidence of neopla-
sia causing dyspnoea as a result of infiltration of the pulmonary
parenchyma or of a neoplastic pleural effusion.
Traumatic dyspnoea was diagnosed based on historical or
clinical evidence of recent severe trauma and findings consistent
with trauma, such as diaphragmatic herniation, radiographic or
postmortem evidence of pulmonary contusions, haemothorax or
pneumothorax.
Cases were excluded if they did not meet the entrance require-
ments, if records were substantially incomplete, if no concerted
attempt was made by the clinician to reach a diagnosis (for
instance, if immediate euthanasia was requested by the owners
before examination could be performed) or if any investigation
was contraindicated on welfare grounds.
Statistical analyses
Statistical analyses were performed using open source and com-
mercially available software (PAST 3, Hammer et al. (2001) and
IBM SPSS Statistics Release 22·0·0·1). When missing data ren-
dered analyses invalid or records were substantially incomplete,
cases were excluded. Alpha was set at 0·05 unless otherwise stated.
Continuous data were assessed for normality graphically and
using the Shapiro–Wilk test. As most data were non-normally
distributed, they were expressed as median value and range. The
difference between groups was examined with a Kruskall–Wallis
test, and where significant differences were identified, pairwise
comparisons were performed using a Mann–Whitney U test
with Bonferroni correction (P<0·008). Categorical data were
expressed as frequencies.
Having established that cardiac dyspnoea was the predomi-
nant diagnosis, we dichotomised the diagnostic categories
Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association
into cardiac and non-cardiac dyspnoea, the latter including all
non-cardiac categories. We then examined whether historical
or clinical findings could reliably differentiate cardiac versus
non-cardiac dyspnoea using a diagnostic test approach. Cases
with witnessed trauma and signs of obvious trauma on physical
examination, such as abrasions, fractures, lacerations or frag-
mented claws, were excluded from the non-cardiac dyspnoea
group under the assumption that clinicians in primary care
practice would attribute dyspnoea to the trauma without fur-
ther initial evaluation. Categorical data were analysed with chi-
squared or Fisher’s exact tests. For the individual temperature/
pulse/respiration (TPR) data, once cases with missing data were
excluded, receiver operating characteristic (ROC) curves were
created. The area under the curve (AUC) for the ROC curve
is a summary statistic of the predictive value across all possible
cut-off points, allowing the classification of a test as highly
accurate (0·9<AUC<1) or moderately accurate (0·7<AUC≤0·9;
Reynolds et al. 2012). Optimal cut-off points were calculated
by optimising for test specificity, and from these, sensitivity,
likelihood ratios and predictive values with confidence intervals
were calculated. Overall test accuracy was calculated as the per-
centage of cases correctly predicted (average of the true positive
and true negative rates).
Logistic regression analyses were performed to evaluate the
association between TPR data and the probability of cardiac dys-
pnoea. Univariable analysis was performed, and variables with
a P value less than 0·20 were subsequently tested in the multi-
variable model. Variables were removed in a backwards stepwise
multi-variable logistic regression model based on a P value greater
than 0·05. We calculated diagnostic test performance, and an
ROC curve was generated to evaluate the model for predicting
whether an animal was in CHF.
Construction of triage algorithm
We created a triage algorithm using clinical examination find-
ings with high test specificity for excluding or diagnosing car-
diac dyspnoea. Each test was run in order, including or excluding
cases with cut-off points optimised to test specificity, dividing the
cases into cardiac or non-cardiac. The diagnostic test performance
of the algorithm was then calculated.
RESULTS
Summary of cases
The entire data set is available in the Figshare link provided in
Data accessibility section. Details of recruiting veterinary prac-
tices and clinicians are given in Table 1. A cohort of 108 cats was
enrolled in this study; details of diagnosis categories are provided
in Table 2; seven were excluded from all analysis: six with sub-
stantially incomplete records and one on treatment for pre-exist-
ing respiratory disease. Of the remaining 101 cases, a definitive
diagnosis was obtained in 92: 60 were cardiac, 15 respiratory, 10
neoplastic and 7 traumatic. In nine cats, no diagnosis was reached;
these remained in the analysis because data collection was com-
plete at presentation.
3
 D. Dickson et al.

Table 1. Supervising, primary clinicians and practices Table 3. Signalment and historical findings in 92 cats
involved in recruiting cases for the RAPID CAT study that were presented with acute dyspnoea in first-opinion
Supervisor Practices Primary Cases practice where a diagnosis was reached
clinicians Cardiac Non-cardiac
1 Barton Veterinary Hospital, 9 60 Respiratory Trauma Neoplasia
Canterbury, Vets Now, Canterbury
n 60 15 7 10
2 Park Vet Group, Cardiff, Gower Vets, 14 44
Swansea, St. James Vets, Swansea, Age, median (min 11·6 (3·5 8 (3 to 18) 3* (2 to 11·8 (7·8
Vale Vets, Dursley to max) to 19) 7·2) to 15·3)
3 St David’s Veterinary Group, Exeter 4 4 Gender (% M) 63 47 28 50
Castle Veterinary Group, Launceston Breed (% Domestic) 93 80 100 90
Torbridge Vet Group, Bideford Median duration of 24 24 1 24
Total 9 27 108 dyspnoea (hours)
Cough 15 (25%) 7 (47%) 1 (14%) 4 (40%)
Supervisor 1 CL, Supervisor 2 DD, Supervisor 3 JH
Wheeze 14 (23%) 7 (47%) 0 3 (30%)
Trauma 0† 0 4 (57%) 0
Weight loss 19 (32%) 6 (40%) 0 4 (40%)
Table 2. Categories of dyspnoea and definitive diagnosis Reduced appetite 38 (63%) 12 (80%) 0 8 (80%)
(when reached) in 108 cats that were presented with Reduced thirst 7 (12%) 1 (7%) 0 3 (30%)
acute dyspnoea in first-opinion practice Increased thirst 7 (12%) 2 (13%) 0 1 (10%)
Diagnosis n % Min to max Minimum to maximum, % M percentage of male, % Domestic Percentage of
domestic breed cats
Cardiac 60 65·2 Cases are grouped by category of dyspnoea aetiology
*Denotes that group median is significantly different from all other groups (Mann–Whitney
Hypertrophic cardiomyopathy 26 43·3
U-test with Bonferonni correction, P<0·008)
Restrictive cardiomyopathy 6 10·0 †Denotes cardiac group is significantly different from non-cardiac group (Fisher’s exact
Non-phenotypic cardiomyopathy 18 30·0 test, P<0·04)
Dilated cardiomyopathy 2 3·3
Arrhythmogenic cardiomyopathy 1 1·7
Aortic endocarditis 1 1·7 with other diagnoses (P<0·008). There were no significant differ-
Hyperthyroid and cardiomyopathy 6 10·0
Respiratory 15 16·3
ences between groups for duration of dyspnoea before presenta-
Pyothorax 7 46·7 tion (P=0·18). When dichotomised into cardiac or non-cardiac,
Asthma 4 26·7 only the historical record of ‘recent trauma’ was different between
Foreign body 1 6·7 groups (P=0·04).
Pneumonia 3 20·0
Traumatic 7 7·6
Clinical examination findings, including TPR, are presented
Diaphragmatic hernia 4 57·1 in Table 4. Only cats with CHF had gallop sounds recognised
Road-traffic accident 3 42·9 by clinicians (14/60; 23·3%; P<0·002). Similarly, cats with
Neoplastic 10 10·9 CHF had lower rectal temperatures and higher respiratory rates
Lymphoma 2 20
Thymoma 2 20
than cats with non-cardiac causes of dyspnoea (P<0·003 for
Carcinoma 3 30 each variable).
Unknown 3 30
Mixed/unknown 9 Temperature, pulse and respiratory rate
Excluded 7
Total 108
From the sample of 101 cats, five cases of traumatic dyspnoea
Category percentages refer to percentages out of diagnosed cases (n=92)
were excluded based on the evidence of witnessed trauma or vis-
ible wounds, leaving a total of 96 cases available for TPR analysis.
Eight cases had missing TPR data and were excluded from the
Six cats with CHF were diagnosed with hyperthyroidism dur- logistic regression modelling (n=88), but were included where
ing hospitalisation, and one cat with CHF due to severe hyper- possible for the individual TPR analyses (n=90 to 96, depending
trophic cardiomyopathy had a solitary lung mass. All these cats on data available for each test). Results of analyses on individual
responded to treatment of CHF and were therefore classified as and combined TPR data are presented in Table 5 and Fig 1, with
cardiac. In the respiratory group, one cat had bronchopneumo- differences in group size for each analysis listed.
nia with concurrent restrictive cardiomyopathy (the latter was Univariable logistic regression identified temperature, heart/
not causing CHF), and another had pyothorax with a concurrent pulse rate and respiratory rate as indicators of CHF. Multi-
thoracic mass (of unknown aetiology). These two cats responded variable logistic regression identified temperature and heart/
to treatment of the respiratory disease; consequently, the pulse rate as independently associated with a diagnosis of CHF.
secondary diagnoses (i.e. restrictive cardiomyopathy and mass, The odds ratio for temperature to predict CHF was 0·293 per
respectively) were considered to be incidental, and so, they were 1°C increase [95% confidence interval (CI): 0·154 to 0·559;
classified as respiratory. P<0·001) and for heart rate was 1·25 per 10 bpm increase
(95% CI: 1·076 to 1·459; P=0·004). The calculated probabil-
Dyspnoea group comparisons ity of CHF using the predicted probabilities derived from the
Signalment and historical findings for each diagnosis group are multi-variable logistic regression gave an ROC AUC of 0·814
summarised in Table 3. Cats with trauma were younger than cats (Table 5).

4 Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association
 RAPID CAT - prospective study of feline dyspnoea

Table 4. Clinical examination findings in 92 cats that were presented with acute dyspnoea in first-opinion practice where
a diagnosis was reached
Cardiac Non-cardiac
Respiratory Trauma Neoplasia
n 60 15 7 10
Mouth breathing 25 (42%) 2 (13%) 2 (29%) 1 (10%)
Abdominal effort 49 (82%) 8 (53%) 5 (71%) 9 (90%)
Pink mmbs 34 (57%) 11 (73%) 3 (43%) 8 (80%)
Pale mmbs 20 (33%) 1 (7%) 2 (29%) 1 (10%)
Blue mmbs 1 (1·7%) 0 0 0
White mmbs 3 (5%) 0 0 0
Red mmbs 1 (1·7%) 1 (7%) 0 0
Murmur 14 (23%) 1 (7%) 0 2 (20%)
Gallop 14* (23%) 0 0 0
Arrhythmia 10 (17%) 0 1 (14%) 0
Inaudible 7 (12%) 1 (7%) 1 (14%) 1 (10%)
Abnormal pulmonary auscultation 47 (78%) 11 (73%) 5 (71%) 4 (40%)
Jugular pulses 4 (7%) 0 0 0
Dull thoracic percussion 5 (8%) 2 (13%) 1 (14%) 2 (20%)
Rectal temperature, median (min to max) 37·8† (35 to 39·8) 38·9† (35·9 to 40·6) 37·8 (37 to 38·5) 38·3 (37·7 to 38·6)
Heart/pulse rate, median (min to max) 192 (90 to 300) 160 (120 to 220) 150 (112 to 220) 180 (140 to 200)
Respiratory rate, median (min to max) 60† (20 to 180) 48† (20 to 68) 44 (24 to 120) 48 (36–70)
mmbs Mucous membranes, min to max Minimum to maximum
Cases are grouped by category of dyspnoea aetiology
*Denotes that cardiac group is significantly different from non-cardiac group (Fisher’s exact test, P<0·002)
†Denotes that group medians are significantly different from each other (Mann–Whitney U test, P<0·003)

Table 5. Diagnostic test performance for discriminating cardiac from non-cardiac dyspnoea based on single and combined
clinical findings in 96 cats that were presented with dyspnoea but without signs of obvious trauma
n Cut-off AUC Sensitivity Specificity PPV NPV +LR −LR Test
accuracy
Rectal 90 <37·5 0·768 44·4% (31 to 94·7% (82 to 92·3% (76 to 54·5% (40 to 8·44 (2·1 to 0·59 (0·46 to 67%
temperature (°C) 59%) 99%) 96%) 91%) 33·6) 0·75)
Heart/pulse rate 91 >192 0·664 52·7% (39 to 78·9% (63 to 78·4% (62 to 53·6% (40 to 2·51 (1·3 to 0·6 (0·43 to 65·3%
66%) 90%) 87%) 74%) 4·9) 0·83)
Respiratory rate 96 >74 0·655 38·3% (26 to 89·5% (75 to 85·2% (67 to 47·9% (34 to 3·64 (1·4 to 0·69 (0·55 to 64·2%
52%) 97%) 91%) 78%) 9·7) 0·87)
*
Logistic regression 88 0·814 77·6% (63 to 71·8% (55 to 77·6% (63 to 71·8% (55 to 2·75 (1·63 to 0·31 (0·18 to 75%
88%) 85%) 88%) 85%) 4·64) 0·54)
†
Triage algorithm 96 — 80% (67 to 89% (73 to 92% (81 to 73% (57 to 7·2 (2·8 to 0·23 (0·14 to 82·5%
89%) 96%) 98%) 85%) 18·3) 0·38)
AUC Area under the curve, PPV Positive predictive value, NPV Negative predictive value, +LR Positive likelihood ratio, −LR Negative likelihood ratio
Values in parenthesis are 95% confidence intervals
*Logistic regression performed using rectal temperature and heart rate as predictor variables, using data from 88 cats
†Triage algorithm, rectal temperature above 40°C to exclude congestive heart failure (CHF), then presence of gallop sound, temperature below 37·5°C, pulse rate greater than 200/minute or
respiratory rate greater than 80/minute to diagnose CHF using data from 96 cats

Using a combination of TPR and gallop sound data, we cre-
ated a triage algorithm to help examine the diagnostic utility of
these data in differentiating cardiac and non-cardiac causes of
dyspnoea (Fig 2). Three cats with a rectal temperature greater
than 40°C had pyothorax; therefore, we used this as the first
step in the algorithm to exclude CHF. Similarly, all 14 cats with
a gallop sound had CHF, effectively removing them from the
questionable subset of cats. We then removed any cats with a
rectal temperature of less than 37·5°C, followed by cats with a
pulse greater than 200 bpm and, finally, the remaining cats with
a respiratory rate greater than 80 per minute as having presump-
tive CHF. The remaining cats were then presumptively classi-
fied as non-cardiac. Diagnostic test performance is reported in
Table 5. Using this algorithm accurately identified 48/60 (80%;
95% CI: 67 to 89%) cats with CHF in the study population.
It also correctly identified 32/36 (89%; 95% CI: 73 to 96%)
cats with non-cardiac dyspnoea. However, this means that 4/36
(11%; 95% CI: 4 to 27%) cats would be incorrectly identified
as having CHF. Using the present approach, such cats would be
at risk of being subjected to inappropriate treatment if no addi-
tional diagnostic testing is performed.
DISCUSSION
We undertook this study to determine the underlying dis-
ease prevalence of acute feline dyspnoea in primary practice in
the UK and to examine the diagnostic utility of a history and

Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association 5
 D. Dickson et al.
FIG 1. Data point distribution plots using (A) temperature (n=90),
(B) heart rate (n=91) and (C) respiratory rate (n=96) to differentiate
between non-cardiac and cardiac aetiology in cats that were
presented with dyspnoea but without signs of obvious trauma.
Hashed line represents the optimal cut-off value for discriminating
between aetiologies calculated from receiver operating characteristic
analysis
6 Journal of Small Animal Practice
initial clinical examination in differentiating causes of dys-
pnoea. Almost two-thirds of the dyspnoeic cats in our study had
CHF, while the remainder had non-cardiac causes of dyspnoea.
Interestingly, very few cats were diagnosed with feline asthma.
We found that simple clinical variables (rectal temperature, heart
rate, respiratory rate and auscultation for a gallop sound) can
provide sufficient information to discriminate between cardiac
and non-cardiac dyspnoea in many instances. In the acute set-
ting, the findings of a low rectal temperature, rapid heart rate,
rapid respiratory rate or a gallop sound should prompt clinicians
to strongly suspect CHF and even institute presumptive therapy
for CHF (specifically diuresis) before additional diagnostic evalu-
ation. The finding of a rectal temperature above 40°C made the
diagnosis of CHF extremely unlikely in our study population.
Cats without these findings require additional diagnostic evalu-
ation before initiating specific treatment. We used these findings
to create a triage algorithm, which might be of use to clinicians
in primary practice. Most historical findings and further clinical
examination did not appear to be useful in the acute setting.
Because the biggest treatment dilemma in dyspnoeic cats for
clinicians lies between treatment for CHF and treatment for
non-cardiac causes of dyspnoea – where therapies are, to a large
extent, diametrically opposed – we dichotomised the diagnostic
categories into cardiac and non-cardiac dyspnoea (which included
all non-cardiac categories). The combination of TPR variables
was ‘moderately accurate’ at discriminating cardiac from non-
cardiac dyspnoea, with a combined test accuracy of 75% to 83%
(Reynolds et al. 2012). Values for temperature, pulse and respira-
tory rate are readily and swiftly acquired in nearly all patients and
require minimal handling or equipment; consequently, the assess-
ment of these clinical variables should be considered mandatory
in any investigation of acute feline dyspnoea. Several other diag-
nostic tests show similar or better diagnostic accuracy in differen-
tiating cardiac from non-cardiac dyspnoea in cats: NT-proBNP
AUC=0·94 (Fox et al. 2009), cardiac troponin-I AUC=0·844 to
0·94 (Herndon et al. 2008, Connolly et al. 2009), thoracic radio-
graphs AUC=0·92 (Sleeper et al. 2013) and echocardiographic
left atrial size AUC=0·91 (Smith & Dukes-McEwan 2012).
However, these tests require patients to be restrained, expertise in
interpretation and prolonged time to perform or obtain results,
which could compromise safety in these physiologically fragile
cats. Additionally, these tests have been validated using referral
practice caseloads or expert evaluations and, often, retrospective
analyses. Therefore, they may be less applicable in an acute pri-
mary practice setting than our findings.
Heart/pulse rate data in dyspnoeic cats in our study showed a
wide spread of values (90 to 300 bpm). This finding corroborates
and extends the work of others who have found that cats with
CHF have widely varying heart rates (Rush et al. 2002, Smith
& Dukes-McEwan 2012). However, we found that the odds of
heart failure increased by about 25% for every 10 bpm increase in
heart rate and that a heart rate greater than 200 bpm was strongly
associated with a diagnosis of CHF in our study population. This
is in contrast to the study by Smith & Dukes-McEwan (2012)
who found that cats with CHF had a lower median heart rate
than cats with subclinical heart disease. However, their study
• © 2017 British Small Animal Veterinary Association
 RAPID CAT - prospective study of feline dyspnoea

FIG 2. Triage algorithm for differentiating cardiac from non-cardiac dyspnoea in 96 cats that were presented acutely in first-opinion practice without
obvious signs of trauma. T Temperature in degrees Celsius, P Pulse, R Respiratory rate, TP True positive, FP False positive, TN True negative, FN False
negative, CHF Congestive heart failure, bpm Beats per minute, br/min Breaths per minute. Number of cats (out of 96) are in circles

included a large proportion of chronic and referred cases, rather
than only those presenting acutely with dyspnoea, and so, their
population is not directly comparable to ours. Tachycardia is not
an unexpected finding in heart failure as these patients would
be likely to have a high sympathetic tone to maintain cardiac
output. The stresses of dyspnoea, travel to a clinic and veterinary
examination will likely exacerbate this. In fact, it is surprising
that many cats present without a tachycardia – i.e. despite severe
dyspnoea (of any aetiology), they often have ‘normal’ heart rates.
In our study, 20 cats had heart rates below 150 bpm. Heart rates
below 150 bpm in the clinic are unusual (Côté 2011b, Smith &
Dukes-McEwan 2012), and the finding of such a heart rate (i.e.
bradycardia) in a cat with a life-threatening condition is difficult
to explain, although Ferasin & DeFrancesco (2015) proposed
that derangements in autonomic tone and accompanying hypo-
thermia may be responsible. Those authors reported that tachy-
cardia is rare in cats with acute CHF, but the veracity of that
statement depends crucially on the definition of tachycardia and
contrasts with our findings.
We also found that a low rectal temperature in cats was
strongly associated with a diagnosis of CHF. The multi-variable
analysis found that for every 1°C decrease in temperature, the
odds of CHF increase by about 3·4 times. Hypothermia might
be a marker of compromised cardiac output or extreme vasocon-
striction in these cats. Peripheral vasoconstriction and hypoten-
sion are inevitable consequences of CHF, and as a result, rectal
temperature may fall. Our findings mirror those of a large ret-
rospective study of hypertrophic cardiomyopathy, in which cats
with CHF or aortic thromboembolism had lower rectal tem-
peratures than those cats presenting with syncope or subclinical
disease (Rush et al. 2002). A study of cats presenting in CHF
by Goutal et al. (2010) found that lower rectal temperatures on
admission was associated with a poorer prognosis, suggesting
that acute life-threatening CHF, with reduced cardiac output,
might result in low rectal temperature and again emphasising
the utility of this simple clinical measurement. Similarly, a recent
review article (Sharp & Rozanski 2013) reported that ‘cats with
CHF are most commonly hypothermic’, although no data were

Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association 7
 D. Dickson et al.
provided to support this statement. Nevertheless, altogether, the
evidence indicates that rectal temperature should be measured in
every dyspnoeic cat provided this can be performed safely.
Conversely, we found that a high rectal temperature in cats
was strongly associated with a diagnosis of non-cardiac disease in
our study population. Three cats exhibited temperatures greater
than 40°C, and all had pyothorax underlying their dyspnoea.
This is not surprising, but it also suggests that a high body tem-
perature should direct the clinician towards treating for non-car-
diac causes or assessing the patient further before implementing
therapy.
The multi-variable model found that respiratory rate did not
contribute significantly and so was removed. While the majority
of cats with a respiratory rate of greater than 74 per minute had
CHF, several cats with non-cardiac dyspnoea had rapid respira-
tory rates. In addition, many of the cats that were tachypnoeic
were also tachycardic, making respiratory rate a non-significant
variable in the multi-variable model. We chose to include respira-
tory rate (>80 breaths per minute) in the triage algorithm because,
when combined with other clinical findings, it improved the test
specificity without overly compromising sensitivity.
Gallop sounds were highly specific for CHF but were infre-
quently observed by clinicians in our study population and clin-
ical setting. Fewer of the cats in our study had gallop sounds
(23%) compared to those reported by Smith & Dukes-McEwan
(2012) or Goutal et al. (2010), who found gallop sounds in
approximately 50% of cats with CHF. We cannot exclude the
possibility that gallop sounds were overlooked by some clini-
cians in some cats. Nevertheless, our data suggest that clinicians
auscultating a gallop sound in a dyspnoeic cat should be highly
suspicious of CHF.
The remaining history and physical examination findings did
not help discriminate between causes of dyspnoea. Many of the
historical questions that were asked (such as changes in appe-
tite or weight) relate to non-specific signs. Murmurs, while most
common in the cardiac group, were not detected in the majority
of CHF patients. Murmurs are an inconsistent finding in cats
with CHF (Rush et al. 2002, Swift et al. 2009, Goutal et al. 2010,
Smith & Dukes-McEwan 2012, Ferasin & DeFrancesco 2015),
possibly due to the difficulty in auscultation in a patient with
tachycardia, pulmonary oedema or pleural effusion. However,
the present data strongly support the view that when manag-
ing an acutely dyspnoeic cat, clinicians should focus on basic
clinical examination and stabilisation of the patient as a priority.
Detailed history taking and thorough physical examination can
then follow.
CHF was the most common cause of sudden onset feline dys-
pnoea in the present study (65% of cats in which a diagnosis
was reached), somewhat higher than that observed by Swift et al.
(2009) (38% of cats) but similar to that observed by Herndon et al.
(2008) (72%), albeit from a referral population. In a recent
study on dyspnoeic animals, performed in a mixture of primary
emergency and referral cardiology practice (82% of cases were
enrolled via primary emergency practice), 62% of dyspnoeic cats
had a diagnosis of CHF (Ward et al. 2017), similar to the preva-
lence in our study. We are all cardiologists, and this may have led
8 Journal of Small Animal Practice
to a bias in the prevalence data (as CHF would be readily diag-
nosed). Similarly, we also acknowledge a possible selection bias in
cases examined because primary clinicians may have been more
likely to ‘recruit’ cases where they suspected a cardiac aetiology
for our study. However, when we briefed the primary clinicians
collecting cases, we did not impose any restrictions on the cases
that could or could not be included, only that they had to present
with dyspnoea as their main clinical sign and that it must be the
first presentation – i.e. they were not referred for investigation
nor did they already have a diagnosis of thoracic disease.
If the prior probability of having CHF as the cause of dys-
pnoea in a cat presenting to a UK primary care practice is high
(e.g. close to 70%), then highly specific clinical findings, such as
low temperature or gallop sounds, can be interpreted and acted
upon with more certainty because false positive findings will be
much less likely. The predictive values derived from our data are
dependent on disease prevalence: in the population we studied,
cardiac dyspnoea was diagnosed most commonly, so tests to
include or exclude cardiac dyspnoea will perform better because
the prevalence is higher. When interpreting our data, clinicians
should be aware of the effects of disease prevalence on test per-
formance and allow for differences in caseload, population and
geography. Furthermore, large-scale studies of disease prevalence
would allow for more accurate test performance statistics, which
is why we consider this study to be preliminary.
A quarter of the cases in our study with CHF had a history of
cough, in direct contrast to a study by Smith & Dukes-McEwan
(2012), which concluded that cats with CHF rarely coughed.
Several other investigators have similarly concluded that cats
with cardiac disease do not cough (Fox 1999, Ware 2007, Corco-
ran 2010, Côté 2011a, Ferasin & DeFrancesco 2015). Our data
do not prove that CHF causes cats to cough, but they do suggest
that a history of cough should not be used to exclude CHF in
cats. It is possible, but unproven, that severe pulmonary oedema
entering the larger airways triggers cough receptors. We did not
ask specifically whether the cough was recent in onset or more
chronic in nature, so it is possible that these cats had a pre-exist-
ing cough, unrelated to the acute dyspnoea and CHF. Similarly, a
cat could be identified as having a history of coughing but might
have had a transient cough, which had resolved by the time of
presentation for dyspnoea.
Somewhat surprisingly, we only diagnosed feline asthma in
4% of cases, despite previous reports suggesting that this is one
of the leading causes of feline dyspnoea (Padrid 2009, Sharp &
Rozanski 2013, Trzil & Reinero 2014). The diagnosis of feline
asthma is sometimes based on responses to corticosteroid therapy,
but corticosteroid therapy is considered contraindicated in CHF
(Rush et al. 2002, Smith et al. 2004). Based on our results, the
prior probability of an acutely dyspnoeic cat in the UK having
asthma at first presentation is low. Therefore, clinicians should
avoid empirical therapy with corticosteroids in an acutely dys-
pnoeic cat if our triage algorithm suggests CHF is the most likely
cause of the dyspnoea.
Following the triage algorithm developed from the present
data, 4/36 cats (11%) could have been misdiagnosed as hav-
ing CHF and been subjected to inappropriate treatment with a
• © 2017 British Small Animal Veterinary Association
 RAPID CAT - prospective study of feline dyspnoea
diuretic had further diagnostic procedures not been performed.
The algorithm we propose should alert a clinician to be suspi-
cious that CHF is causing the dyspnoea and to potentially insti-
tute emergency therapy, but it should not be the sole diagnostic
tool. As soon as it is safe and practical to do so, confirmation
of the underlying diagnosis should be sought. The overwhelm-
ing majority of cats in this study were investigated at an early
stage, at least briefly, using thoracic ultrasound, facilitating the
swift recognition of pleural effusions or left atrial dilation. Thus,
cardiac disease and CHF, as well as other important causes of
dyspnoea, including pyothorax and diaphragmatic hernias, were
rapidly diagnosed. In an earlier study of cats in primary care
practice performed by a cardiologist, left atrial to aortic ratio
was a very useful diagnostic test to confirm CHF (AUC=0·91;
Smith & Dukes-McEwan 2012). We believe that the safest strat-
egy with the current state of knowledge is to perform a basic
focused clinical examination and then put the cat into an oxy-
gen-enriched environment without further testing, at least until
the respiratory condition appears to have improved. If the basic
examination findings as detailed above suggest that the dyspnoea
is of cardiac origin, then echocardiography, if available, should
be performed to confirm the diagnosis, and a diuretic should be
administered as early as is practical. In a case where clinical find-
ings are non-discriminatory, further diagnostic testing would be
required before starting drug treatment. Clinicians should avoid
procedures that require considerable manual restraint (such as
taking a lateral radiograph or venepuncture) until it appears safe
to do so.
Although the primary care practice perspective of the current
study can be regarded as a strength, there are some associated
limitations. It was performed without research funding, and all
tests were consequently performed at the owners’ expense, lim-
iting the investigation in some cases. It is possible that some
cats were presented with multiple aetiologies of which we were
unaware. However, in most cases, we found convincing evidence
of a primary disease process, which was supported by appropri-
ate diagnostic testing and response to treatment or confirmation
at time of death. Dyspnoea is a subjective clinical sign with a
variable presentation, and therefore, we may have excluded cases
where dyspnoea was subtle or other more prominent clinical
signs were identified (such as a fractured limb or paraparesis)
that meant the dyspnoea was less relevant. However, our study
was designed specifically to address the presentation of dyspnoea
in a cat without preceding signs, which often presents a chal-
lenge to clinicians.
Despite conducting a prospective study over a 5-year period,
sample sizes in the non-cardiac groups were relatively small,
which could influence the conclusions we have drawn. In addi-
tion to an increased number of subjects and more standardised
confirmatory testing, enhancements to future studies in this area
would likely include recording a diagnosis of pulmonary oedema
versus pleural effusion in CHF cases. It would also be useful to
obtain data on certain other clinical findings, such as jugular pul-
sation and thoracic percussion resonance.
In conclusion, heart failure appears to be the most common
cause of acute dyspnoea in cats that are presented to first-opinion
Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association
practitioners in the UK. The present study indicates that recording
and assessing basic clinical findings (especially temperature, pulse
and respiratory rate) in cats that are presented with acute dyspnoea
provides valuable diagnostic information that can be used with
some confidence in triage to differentiate cardiac from non-cardiac
dyspnoea in non-trauma cases.
Acknowledgements
We thank Professor Adrian Boswood for statistical advice,
Dr. Mark Patteson for guidance and support provided during
the study period and Dr. Andy Wales for detailed and criti-
cal manuscript review. We also gratefully acknowledge all the
primary clinicians and practices who were involved with this
study.
Conflict of interest
The authors have no conflicts of interest to disclose. The study
was not supported by any grant or funding agency.
References
Abbott, J. A. & MacLean, H. N. (2006) Two-dimensional echocardiographic
assessment of the feline left atrium. Journal of Veterinary Internal Medicine
20, 111-119
Connolly, D. J., Copel, H., Collins, S., et al. (2009) Assessment of the diagnostic
accuracy of circulating cardiac troponin I concentration to distinguish between
cats with cardiac and non-cardiac causes of respiratory distress. Journal of
Veterinary Cardiology 11, 71-78
Corcoran, B. (2010) Clinical approach to coughing. In: Manual of Canine and
Feline Cardiorespiratory Medicine. 2nd edn. Eds V. L. Fuentes, L. R. Johnson
and S. Dennis. BSAVA, Gloucester, UK. pp 11-14
Côté, E. (2011a) Dyspnoea and cough. In: Feline Cardiology. Eds E. Côté, K.
MacDonald, K. Meurs and M. Sleeper. Wiley and Sons, Chichester, UK. pp
25-27
Côté, E. (2011b) Arrhythmias. In: Feline Cardiology. Eds E. Côté, K. MacDonald, K.
Meurs and M. Sleeper. Wiley and Sons, Chichester, UK. p 216
Ferasin, L. & DeFrancesco, T. (2015) Management of acute heart failure in cats.
Journal of Veterinary Cardiology 17, S173-S189
Fox, P. (1999) The history. In: Textbook of Canine and Feline Cardiology. Eds P.
R. Fox, D. Sisson and N. S. Moise. W. B. Saunders, Philadelphia, PA, USA. pp
41-45
Fox, P. R., Oyama, M. A., Reynolds, C., et al. (2009) Utility of plasma N-terminal
pro-brain natriuretic peptide (NT-proBNP) to distinguish between congestive
heart failure and non-cardiac causes of acute dyspnea in cats. Journal of Vet-
erinary Cardiology 11, S51-S61
Goutal, C. M., Keir, I., Kenney, S., et al. (2010) Evaluation of acute congestive
heart failure in dogs and cats: 145 cases (2007-2008). Journal of Veterinary
Emergency and Critical Care 20, 330-337
Hammer, Ø., Harper, D. & Ryan, P. D. (2001) PAST: paleontological statistics soft-
ware package for education and data. Palaeontologia Electronica 4, 9
Herndon, W. E., Rishniw, M., Schrope, D., et al. (2008) Assessment of plasma
cardiac troponin I concentration as a means to differentiate cardiac and non-
cardiac causes of dyspnea in cats. Journal of the American Veterinary Medical
Association 233, 1261-1264
Kellett-Gregory, L. M. & King, L. G. (2010) Management of acute respiratory dis-
tress. In: Manual of Canine and Feline Cardiorespiratory Medicine. 2nd edn.
Eds V. L. Fuentes, L. R. Johnson and S. Dennis. BSAVA, Gloucester, UK. pp
142-152
Mandell, D. C. (2004) Respiratory distress in cats. In: Textbook of Respiratory
Disease in Dogs and Cats. Ed L. G. King. Saunders, St. Louis, MO, USA. pp
12-17
Padrid, P. (2009) Chronic bronchitis and asthma in cats. In: Current Veterinary
Therapy XIV. Eds J. Bonagura and D. Twedt. Saunders, Philadelphia, PA, USA.
pp 650-658
Reynolds, C. A., Brown, D. C., Rush, J. E., et al. (2012) Prediction of first onset
of congestive heart failure in dogs with degenerative mitral valve disease: the
PREDICT cohort study. Journal of Veterinary Cardiology 14, 193-202
Rush, J. E., Freeman, L. M., Fenollosa, N. K., et al. (2002) Population and
sur vival characteristics of cats with hyper trophic cardiomyopathy: 260 cases
(1990-1999). Journal of the American Veterinar y Medical Association 220,
202-207
Sharp, C. R. & Rozanski, E. A. (2013) Physical examination of the respiratory
system. Topics in Companion Animal Medicine 28, 79-85
9
 D. Dickson et al.
Sleeper, M. M., Roland, R. & Drobatz, K. J. (2013) Use of the vertebral heart scale
for differentiation of cardiac and noncardiac causes of respiratory distress in
cats: 67 cases (2002-2003). Journal of the American Veterinary Medical Asso-
ciation 242, 366-371
Smith, S. & Dukes-McEwan, J. (2012) Clinical signs and left atrial size in cats
with cardiovascular disease in general practice. The Journal of Small Animal
Practice 53, 27-33
Smith, S., Tobias, A., Fine, D., et al. (2004) Corticosteroid-associated congestive
heart failure in 12 cats. International Journal of Applied Research in Veterinary
Medicine 2, 1-12
10
Swift, S., Dukes-McEwan, J., Fonfara, S., et al. (2009) Aetiology and outcome in
90 cats presenting with dyspnoea in a referral population. Journal of Small
Animal Practice 50, 466-473
Trzil, J. E. & Reinero, C. R. (2014) Update on feline asthma. Veterinary Clinics of
North America: Small Animal Practice 44, 91-105
Ward, J. L., Lisciandro, G. R., Keene, B. W., et al. (2017) Accuracy of point-of-care lung ultra-
sonography for the diagnosis of cardiogenic pulmonary edema in dogs and cats with
acute dyspnea. Journal of the American Veterinary Medical Association 250, 666-675
Ware, W. A. (2007) Cardiovascular Disease in Small Animal Medicine. Manson,
London, UK. pp 101-105
Journal of Small Animal Practice • © 2017 British Small Animal Veterinary Association