The accuracy of tactile assessment of canine nose temperature to identify rectal hyperthermia and hypothermia in dogs presenting on an emergency basis

Abstract

The goals of this study were to evaluate whether touch can identify a warm nose as opposed to a cold nose, to examine the correlation between thermographically measured nose temperatures and rectal temperatures, and to calculate the accuracy of tactile assessment of nose temperature in detecting rectal hyperthermia and hypothermia in dogs. A total of 100 dogs presenting to an emergency room was prospectively enrolled. Tactile nose assessment was carried out on triage. Noses were subjectively categorized as warm, cold, or intermediate (neither warm nor cold). Thermographic nose temperatures were recorded using a thermal imaging camera. Tactile assessment categorized noses as warm, intermediate, or cold (P < 0.01). There was no correlation between thermographically measured nose temperature and rectal temperature (r = 0.02). Tactile assessment of noses as warm had a sensitivity of 29.4% and a specificity of 79.5% for detecting rectal hyperthermia; calculated test accuracy was 71%. Tactile assessment of noses as cold had a sensitivity of 54.5% and a specificity of 62.9%; calculated test accuracy was 62%. It was concluded that nose temperatures do not correlate with rectal temperatures. Tactile assessment of nose temperature is inaccurate for identifying rectal hyperthermia or hypothermia.

Introduction

Clients may believe that nose temperature is a way of assessing their dog’s health (1). Dog owners may become concerned with apparent changes in the nose temperature, which are then often included as a presenting complaint in patient history.

Prior to the onset of panting in dogs, evaporative heat loss occurs due to increased blood flow, i.e., convective heat transfer, to the nose (2). Additionally, the canine nose is known to play a role in regulating brain temperature (3). It would therefore not be unreasonable to think that the nose represents thermal status in dogs.

A systematic review in humans concluded that a mother’s touch had a sensitivity of 90% and a specificity of 50% for detecting fever in children, which indicates that mothers were good at detecting fever in febrile children, although they may overestimate fever in a normal child (4). These findings were supported by a more recent systematic review and meta-analysis in children, which suggested that a tactile assessment of “not febrile” was reasonably accurate, whereas “febrile” may represent a false positive (5).

Rectal temperature remains the accepted standard: neither temporal nor ear thermometers accurately assess temperature in children, although they may be useful as screening techniques (6). While Southward et al (7) reported strong correlation between auricular and rectal temperatures in dogs, proxy site temperature measurements, i.e., axillary, auricular, corneal, and subcutaneous, should similarly be considered at best as screening techniques (8–11).

The correlation between canine nose temperature and rectal temperature is unknown. The ability to identify rectal hyperthermia and hypothermia in dogs by tactile assessment of the nose is also unknown. The hypotheses of this study were: i) cold noses feel cold and warm noses feel warm; ii) nasal temperature correlates with rectal temperature; and iii) tactile assessment of canine nose temperature can identify rectal hyperthermia and hypothermia.

Materials and methods

A convenience sample of 100 dogs presenting on an emergency basis from January 1, 2019 and May 1, 2019 were enrolled. Dogs were eligible for enrolment if they had an intact nose and were amenable to nasal palpation and thermal imaging, i.e., could be photographed. Dogs were assessed during triage; there was no regulated period of acclimatization to the hospital’s ambient temperature.

Patient details, including age, sex, breed, and weight, and triage vitals, including rectal temperature, were obtained from the patient records. Dogs were categorized as having rectal hyperthermia (> 39.3°C) or hypothermia (< 37.8°C) (12). Patients were excluded if any of the required data were missing from their records.

A single assessor (CK) carried out tactile and thermographic assessment of each dog’s nose, without objective knowledge of vital parameters or patient stability. Tactile assessment was carried out first. Following routine hand hygiene with an alcohol-based hand sanitizer, the palmar aspects of the 2nd and 3rd digits of the left hand were held in gentle contact with the anterior-most aspect of each dog’s nose for 15 s. Noses were then categorized as warm, intermediate, or cold based on subjective assessment only. Noses were classified as “intermediate” when the assessor did not feel that a nose felt convincingly warm or cold.

A thermal imaging camera (FLIR ONE Gen 3; FLIR Systems, Richmond, British Columbia) was used to objectively measure nose temperatures in degrees Celsius (Figures 1, 2). The camera was held approximately 50 to 100 cm away from the dog’s nose. The focal-point cursor of the camera’s thermometer was centered on the anterior-most aspect of the nose, typically the philtrum.

Figure 1.

Thermographic image of (A) a dog with a cold nose and (B) a dog with an intermediate nose.

Figure 2.

Thermographic image of (A) a dog with a warm nose and (B) a dog with a cold nose.

Continuous data were tested for normality with the Shapiro-Wilk test. Correlation between nonparametric data was evaluated with Spearman’s rank correlation coefficient. Categorical and binary variables were reported as a proportion. When comparing groups of more than 2 categorical variables with nonparametric continuous data, a Kruskal-Wallis H-test was used.

The sensitivity and specificity of tactile assessment of temperature for identifying rectal hyperthermia and hypothermia were calculated. Positive predictive value (PPV), negative predictive value (NPV), and accuracy of tactile assessment for identifying rectal hyperthermia and hypothermia were also calculated.

The Institutional Animal Use and Care Committee waived client consent for this study due to the non-invasive nature of the thermal imaging.

Results

A total of 100 dogs was enrolled in this study. A variety of breeds was represented: 24 mixed breeds; 16 Labrador retrievers; 5 golden retrievers, 5 German shepherds; 3 boxers, 3 English bulldogs, 3 pit bulls, and 3 standard poodles; the remainder were various other pedigree breeds. The median age was 9 y (range: 0.5 to 17.5 y) and the median weight was 20.55 kg (range: 3 to 57.9 kg).

The data for rectal temperature were not normally distributed (P = 0.003). Seventeen dogs (17%) presented with rectal hyperthermia (39.4 to 40.8°C). Eleven dogs (11%) presented with rectal hypothermia (35 to 37.7°C).

The data for thermographic nose temperature were not normally distributed (P = 0.023). Median for these data was 25.15°C (range: 16 to 38.3°C). Based on tactile assessment, 22 noses (22%) were categorized as warm, 39 (39%) were categorized as intermediate, and 39 (39%) were categorized as cold. The median of thermographic temperatures was 31.5°C (range: 25.2 to 38.3°C) for noses categorized as warm, 25.7°C (range: 17.8 to 34.6°C) for noses categorized as intermediate, and 21.0°C (range: 16.0 to 26.8°C) for noses categorized as cold (Figure 3). Nose temperatures were significantly different (P < 0.001) among tactile groups.

Figure 3.

Box plot showing the distribution of thermographic nose temperatures categorized into the 3 tactile assessment groups.

There was no correlation between thermographic nose temperature and rectal temperature (r = 0.02) (Figure 4). Based on tactile assessment, dogs categorized as having a warm nose had a median rectal temperature of 38.65°C (range: 35 to 39.9°C), dogs with an intermediate nose had a median rectal temperature of 38.7°C (range: 37.7 to 40.6°C), and dogs with a cold nose had a median rectal temperature of 38.7°C (range: 36.6 to 40.8°C) (Figure 5).

Figure 4.

Scatter plot showing thermographic nose temperature versus rectal temperature. The dotted lines show the upper and lower limits of normal rectal temperature. The solid line shows best-fitting linear correlation.

Figure 5.

Box plot showing the distribution of rectal temperatures within the 3 tactile temperature assessment groups.

Tactile assessment of nose temperature as warm had a sensitivity of 29.4% and a specificity of 79.5% for detecting rectal hyperthermia [positive predictive value (PPV) = 22.7% and negative predictive value (NPV) = 84.6% in this population]. The calculated accuracy of tactile assessment of nose temperature to correctly identify whether or not a dog had rectal hyperthermia was 71%. Tactile assessment of nose temperature as cold had a sensitivity of 54.5% and a specificity of 62.9% for detecting rectal hypothermia (PPV = 15.4% and NPV = 91.8% in this population). The calculated accuracy of tactile assessment of nose temperature to correctly identify whether or not a dog had rectal hypothermia was 62%.

Discussion

This study evaluated relationships between both tactile assessment and thermographic measurement of nose temperature and rectal temperature in dogs. Tactile assessment, or the sensory assessment of noses by touch, correctly grouped noses into warm, intermediate, and cold, but did not identify rectal hyperthermia or hypothermia. Objective temperature measurement with a thermographic camera, the FLIR ONE Gen 3, found no correlation between measured nose temperature and rectal temperature.

Tactile assessment of nose temperature was able to differentiate warm, intermediate, and cold noses as measured by thermography. This suggests that owners may similarly assess their dogs’ noses using touch. Despite statistical differences, the ranges overlapped so that noses could therefore be incorrectly classified as warm, intermediate, or cold (Figure 3).

Nose temperature recorded thermographically did not correlate with rectal temperature in this study and cannot be considered a substitute for measuring rectal temperature (Figure 4).

Noses categorized as warm by tactile assessment had a very low sensitivity (29.4%) and a moderate specificity (79.5%) for detecting rectal hyperthermia. This suggests that the ability to detect rectal hyperthermia in dogs by tactile assessment of the nose temperature is very poor. It may be reasonable, however, to use tactile assessment to infer the absence of rectal hyperthermia. This differs from findings in human studies that a mother’s touch is sensitive enough to detect a fever in their child but may infer fever when it is not actually present, i.e., low specificity (4,5). The moderate NPV of 84.6% indicates a reasonable chance that a nose not feeling warm correctly identifies the absence of rectal hyperthermia in this population, although this needs to be validated in a larger data set.

It is important to note that the low number of dogs with rectal hyperthermia in this study population (n = 17) risks biasing the data in favor of a higher NPV. The low calculated accuracy (71%) of determining the presence versus absence of rectal hyperthermia based on tactile assessment suggests that this is not a useful test.

Noses categorized as cold by tactile assessment had a low sensitivity (54.5%) and a low specificity (62.9%) for detecting rectal hypothermia. This suggests that the ability of tactile assessment to infer the presence or absence of hypothermia is low. The high NPV (91.8%) for correctly identifying a dog that did not have rectal hypothermia is valid only in this study population and is also influenced by the low number of dogs with rectal hypothermia (n = 11). Tactile assessment is therefore not useful for detecting rectal hypothermia in dogs due to its low accuracy (62%).

It was concluded that, while tactile assessment can differentiate between warm, intermediate, and cold noses, it cannot be used to predict rectal temperature. Thermographic measurement of nose temperature did not correlate with rectal temperature and cannot be used to approximate rectal temperatures. This study was conducted in a hospital environment in which rectal temperature measurement was readily available. Although often reported as part of the medical history, these data indicate that tactile assessment of nose temperature should not be considered a sensitive or specific indicator of temperature extremes in dogs.