Abstract
Objective
To study the feasibility and test–retest repeatability of a sensory threshold examination protocol (STEP) and report the quantitative sensory threshold distributions in healthy dogs.
Study design
Prospective, observational, cohort study.
Animals
Twenty-five healthy client-owned dogs.
Methods
Tactile sensitivity test (TST) (von Frey filaments), mechanical thresholds (MT with 2, 4 and 8 mm probes), heat thresholds (HT) and responsiveness to cold stimulus (CT at 0 °C) were quantitatively assessed for five body areas (BAs; tibias, humeri, neck, thoracolumbar region and abdomen) in a randomized order on three different occasions. Linear mixed model and generalized linear mixed models were used to evaluate the effects of body weight category, age, sex, BA, occasion, feasibility score and investigator experience. Test–retest repeatability was evaluated with the intra-class correlation coefficient.
Results
The STEP lasted 90 minutes without side effects. The BA affected most tests (p ≤ 0.001). Higher thresholds and longer cold latencies were scored in the neck (p ≤ 0.024) compared to other BAs. Weight category affected all thresholds (p ≤ 0.037). Small dogs had lower MT (∼1.4 N mean difference) and HT (1.1 °C mean difference) than other dogs (p ≤ 0.029). Young dogs had higher HT than adults (2.2 °C mean difference) (p = 0.035). Gender also affected TST, MT and HT (p < 0.05) (females versus males: TST odds ratio = 0.5, MT = 1.3 N mean difference, HT = 2.2 °C mean difference). Repeatability was substantial to moderate for all tests, but poor for TST. There was no difference in thresholds between occasions, except for CT. Test–retest repeatability was slightly better with the 2 mm MT probe compared to other diameters and improved with operator experience.
Conclusions and clinical relevance
The STEP was feasible, was well tolerated and showed substantial test–retest repeatability in healthy dogs. Further validation is needed in dogs suffering pain.
Keywords
Introduction
Quantitative sensory testing (QST) is a method used to quantify the somatosensory function (
Backonja et al., 2013
, Edwards et al., 2016
). In the clinical research setting, QST evaluation comprises touch and vibration detection, as well as mechanical and thermal (heat and cold) noxious stimuli (Walk et al., 2009
). In humans, QST has been applied in healthy volunteers, patients with neuropathic pain defined as ‘pain caused by a lesion or disease of the somatosensory nervous system’ (Backonja et al., 2013
, Finnerup et al., 2016
) and other pain syndromes where the somatosensory function may be altered as a result of peripheral or central sensitization (Whitaker et al., 2016
).There is evidence in people that altered somatosensory function originates from various pathophysiological mechanisms that can be elucidated by the results of a QST panel (
Greenspan, 2001
, Hansson, 2002
, Hansson et al., 2007
). The QST may identify patient subgroups with certain underlying neurobiological mechanisms who may respond differently to a given drug (Baron and Dickenson, 2014
). Characterizing the somatosensory phenotype of patients with chronic pain by identifying sensory abnormalities (positive, such as hyperalgesia and allodynia, or negative, such as numbness or lack of sensation) is necessary to help select the best therapeutic class for a specific patient. This is the key to mechanism-based diagnosis and could significantly improve treatment (Rolke et al., 2006
, Reimer et al., 2014
, Edwards et al., 2016
).Similar to humans, animals experience chronic pain of neuropathic origin (
Mathews, 2008
). The QST has the potential to be a neurophysiological tool in veterinary medicine and has been used in different clinical and experimental models such as osteoarthritis, hip replacement and ovariohysterectomy in dogs (Brydges et al., 2012
, Hunt et al., 2013
, Moore et al., 2013
, Tomas et al., 2014
). Recently, thermal stimuli have been tested in combination with mechanical stimuli in canine models of osteoarthritis and spinal cord injury (Gorney et al., 2016
, Knazovicky et al., 2016
, Song et al., 2016
). However, the combination of all the QST modalities together in one standardized test has never been explored.The use of naturally occurring canine pain models is becoming a valuable option to study human chronic pain (
Lascelles, 2013
). They better mirror human conditions and may provide better insight into drug efficacy in humans compared with experimentally induced rodent models. Observing the responses of dogs administered analgesic drugs for different naturally occurring pathophysiologic mechanisms offers powerful models for translational studies. Designing a standardized method to evaluate nociceptive thresholds in canine patients and defining sources of confounding factors in healthy dogs will ultimately offer an improvement of diagnosis and characterization of chronic pain.The aims of this study were to: 1) evaluate the feasibility and test–retest repeatability of a QST sensory threshold examination protocol (STEP) including tactile, thermal and mechanical testing; 2) identify explanatory variables affecting results; and 3) provide baseline QST thresholds and their distribution in a sample of healthy dogs for its use as a tool to phenotype chronic pain syndromes in future studies.
Materials and methods
The project was approved by the Royal Veterinary College Ethics and Welfare Committee (URN 2013 1243). Twenty-five healthy client-owned dogs were included in the study which was conducted between January and August 2014. Signed owner consent was obtained for all animals enrolled in the study. The dogs were deemed healthy based on their medical history and a complete physical/neurological/orthopaedic examination performed by a veterinarian. Owners completed the Canine Brief Pain Inventory (CBPI), which consisted of three parts: pain severity ranging from 0 (no pain) to 10 (extreme pain), pain interference from 0 (no interference) to 10 (completely interferes) and quality of life assessment from 1 (poor) to 5 (excellent) (
Brown et al., 2008
). An inclusion criterion was a CBPI score of 0 on pain severity and pain interference, with a quality of life scores greater than 4 (very good) (Brown et al., 2008
). Dogs that were not able to attend a minimum of two appointments (occasions) were excluded.Animals were tested on two or three occasions (occasion 1, 2 or 3) (Fig. 1), each separated by a week, with a STEP. The CBPI was completed on each occasion to ensure that no changes occurred over time in order to continue in the study. The standardized STEP consisted of a tactile sensitivity test (TST using von Frey filaments), mechanical thresholds (MT using a calibrated veterinary pressure algometer), heat and cold thresholds (HT, CT). The tests were applied in the same order in all dogs as follows: TST, MT, HT, CT. Mechanical testing was performed before thermal testing to avoid iatrogenic sensitization, according to
Grone et al., 2012
. For each sensory modality, measurements were taken from five different body areas (BAs) in a randomized order (www.graphpad.com/quickcalcs): bilaterally over the mid tibias, mid humeri, neck area, thoraco-lumbar area and left side only over the abdomen (Fig. 2). Dogs were all tested in the same room in standing position. Prior to testing, dogs were acclimatized to the room for 5 minutes before clipping. Clipping of the BAs (1.5 × 1.5 cm patch) was needed to allow TST and thermal evaluation. The areas were clipped on each occasion. The test started not less than 10 minutes after clipping.
Figure 1Consort flow diagram of dogs included in the study. TSTUD, up–down technique method of testing; TST50%, 50% of response technique method of testing; MT, mechanical threshold 2, 4 and 8 mm size probe; HT, heat threshold; CT, cold threshold.
Figure 2Body areas tested and anatomical localization. 1) Left and right tibias: mid-point between the stifle joint and the hock on the lateral aspect of the tibia. 2) Left and right humeri: mid-point between the scapulo-humeral joint and the elbow on the lateral aspect of the humerus. 3) Left and right neck: mid-point between the atlas wings and the cranial aspect of the body of the scapula on the lateral aspect of the neck. 4) Left and right thoraco-lumbar (T-L): palpate the last rib-vertebrae union. At that level, palpate the spinous process. Testing point is located 1 cm (small dog) to 3 cm (large dog) lateral to the spinal process. 5) Left abdomen: mid-point between midline and the fold of the flank. Illustration courtesy of Mrs Carol Hoy.
Each individual test terminated with the observation of one of the following endpoints: turning the head towards the device, growling, lip licking, or backing away from the stimulus. A feasibility score ranging from 1 (no problem) to 5 (impossible) adapted from
Briley et al., 2014
(Appendix 1) was used to evaluate dog cooperation. All of the tests were readily escapable and, if an animal appeared to be in discomfort during testing (or unable to tolerate the protocol), the test was terminated immediately. If the dogs showed fatigue or reluctance to stand, time was allowed for resting of up to 5 minutes between tests.TST thresholds
Von Frey filaments [20 filaments, 0.008 to 300 gram force (gf); Bioseb, France] were used for TST. The hairs were pressed against the skin with enough force so that the hair buckled and formed a U shape. Two techniques were applied and compared. First, a group of 18 dogs were tested with the ‘up–down technique’ (TSTUD) described by
Chaplan et al., 1994
. The test was initiated with an intermediate 2.0 gf hair. A lack of response to a filament dictated that the next thickest filament was used in the following stimulation (‘up rule’), whereas a positive response dictated the use of the next thinnest filament (‘down rule’). When the animal first changed its response pattern: a negative response followed by a positive response or vice versa, another four von Frey presentations were done according to the above ‘up–down rules’. The final response threshold was interpolated using the following formula: gf threshold = (10 [Xf + kδ])/10,000, where Xf is the value (in log units) of the final von Frey filament used, k is the tabular value (see Chaplan et al., 1994
for more details) for the pattern of positive/negative responses, and δ is the mean difference (in log units) between stimuli.The 50% response technique (TST50%) described by
Brydges et al., 2012
was used in a second group of seven dogs, because preliminary data from the up–down technique suggested difficulties in interpretation of the final threshold as a result of data censoring (animals not responding to the thickest filament). The TST50% consisted of using the filaments in ascending order. Each filament was applied six times, with 3 second intervals. If no aversive response was obtained after testing with a small diameter filament, the next highest diameter filament was used. The TST was defined by the filament that first induced a withdrawal response at least three times in six repeated measurements.Mechanical thresholds
Mechanical response was tested with a calibrated veterinary pressure algometer (ProdPro; Topcat Metrology Ltd, UK), equipped with three different probe diameters: 2, 4 and 8 mm. The accuracy of the instrument was ± 0.5 Newton (N) within a range of 0.5–25 N. The algometer provided a constant increment pressure increase of 2 N second−1 to achieve repeatable applications. The device was applied perpendicular to the skin of the dogs with one hand. The other hand was used to support gently the medial aspect or the contralateral side of the area tested. Three repetitions in the five BAs were obtained for each occasion with the three different probe sizes. Twenty seconds was allowed between repetitions. The final thresholds for the occasion were obtained calculating the mean of the three repeats per BA.
Thermal thresholds
Heat stimulus was applied using a veterinary thermal probe (HotPro; Topcat Metrology Ltd). The device was a handheld calibrated prototype adapted from the already validated wired version (
Dixon et al. 2002
). Before testing, the skin temperature was measured with the device and room temperature was recorded (EL-USB-TP-LCD; Lascar Electronics, UK). During testing, the temperature increased from baseline to a maximum of 55 °C with a ramp of 1 °C second−1 until the endpoint was reached. The device was applied as described in the use of the pressure algometer. Three repetitions in the five BAs were obtained for each occasion. Twenty seconds was allowed between repetitions. The final threshold for the same occasion was obtained calculating the mean of the three repeats per BA.Cold stimulus was applied using a handheld thermal probe (NTE-2A; Physitemp Instruments, NJ, USA) with a 13 mm diameter surface set at 0.0 ± 0.2 °C. The probe used a peltier semiconductor heat pump and a digital temperature control unit to maintain accurate temperature application during trials. The latency (seconds) between application and observation of endpoint was recorded. Three repetitions for each BA were obtained on each occasion. Each repeat included the entire series of BAs in a randomized order, starting again the entire series in the same random order for the second and the third repeat. This allowed at least 60 seconds between repeats in the same BA, maintaining appropriate duration of the total time spent in all the tests.
Analysis of data
Data were analysed using statistical software (SPSS 21, IBM, Portsmouth, UK). Data from dogs in which the feasibility scores were higher than 2 were excluded from the analysis. For continuous data, normality of distribution was verified by Kolmorov–Smirnov's test and by visual assessment of Q–Q plots and histograms. When required, data were logarithmically transformed to verify the assumption of data normality prior to parametric testing. Cold and tactile sensitivity thresholds were right-censored (60 seconds and highest filament, respectively) and treated as binary data (0 = response below threshold and 1 = threshold reached).
Continuous data were expressed as mean ± standard deviation (SD). Data following a logarithmic distribution were presented as geometric mean and back-transformed SD. Other data were presented as median (range). For graphical display, median, interquartile range and minimum–maximum was used. Categorical data were expressed as number out of total and percentage. Significant differences were considered if p < 0.05.
Data were divided into two periods of testing (first period of testing from January 2014 to April 2014 against the second period of testing from May 2014 to August 2014) to evaluate the effect of the operator gaining experience with QST thresholds.
A linear mixed model was used for continuous outcome variables MT (N) and HT (°C) separately, to evaluate the influence of the explanatory variables on within-/between-subject variability. Subjects were considered as a random effect. The following explanatory variables were considered as fixed effects: body weight and age (divided in three categories respectively, Appendix 2), sex, BA (5 total), right/left side. Analysis of HT also included body temperature and room temperature as additional fixed effects. Factors affecting the metrological performance of the protocol were also included in the model as fixed effects: feasibility score (0, 1 or 2), effect of repeated testing (occasion 1, 2 or 3) and period of testing (first and second periods). In the case of the pressure algometer, the three different probes (2, 4 and 8 mm) were compared in separate statistical models (MT2, MT4, MT8). The magnitude of the effects was reported as the adjusted mean difference and p value.
A generalized linear mixed model was used for tests with binary logistic outcomes (TST and CT). The dependent variables were response to any of the von Frey filaments and 0 °C before 60 seconds (pTST and pCT), respectively. The fixed and random effects were the same as for continuous outcomes. The magnitude of the effects was reported as the odds ratio and p value.
Interactions were evaluated when appropriate. Post hoc comparisons of the significant effects were made using Fisher's least standard differences method.
Test–retest repeatability was evaluated by calculating the intra-class correlation coefficient (ICC). The ICC is the degree of closeness of repeated measures in a group of individuals (
Andersen et al., 2014
). It describes the contribution of the variation within the individual within the total variation (between dogs variation + within dogs variation + error variation) (Vangeneugden et al., 2004
). Therefore, the closest to 1 the ICC, the smallest the variation within dogs across the different occasions (occasion 1, 2 or 3), and the better the repeatability of the test. The ICCs were categorized as slight/poor (<0.2), fair (>0.2 to 0.4), moderate (>0.4 to 0.6), substantial (>0.6 to 0.8) and almost perfect (>0.8) (Landis and Koch, 1977
).Results
Descriptive results
The 25 healthy client-owned dogs included in the study (Fig. 1) had an age of 6.0 (0.3–9.0) years and a body weight of 15 (6–35) kg. There were 14 females (56%) and 11 males (44%). All dogs' CBPI scores were 0 for pain intensity and pain interference, and 5 for quality of life. Eleven dogs (44%) were tested during the first period of testing. Distributions of the sample by different weight category and age are shown in Appendix 2. Feasibility score distribution across the sample of dogs was 0 for four dogs, (16%); 1 for nine dogs, (36%); 2 for 12 dogs, (48%). The temperature of the testing room was 22.9 (19.3–26.2) °C. The skin temperature was 30.9 (27.6–33.2) °C. According to this range of skin temperature, the baseline starting temperature was set at 30 °C for HT in all dogs. The STEP protocol took 90 minutes per dog and was applied with no side effects reported by owners.
Mean ± SD or median (range) of the TST, MT, HT, and CT are displayed in Table 1, Table 2, respectively. Median (interquartile range) and minimum–maximum thresholds for the different stimuli are summarized for the different BAs in Fig. 3.
Table 1Mechanical (MT) and heat thresholds, mean, standard deviation (SD) and range obtained for the different probes and the different weight categories. Response to tactile stimulus and cold stimulus (%), tactile sensitivity threshold (TST) method 1 and 2 and cold latency (at 0 °C), median and range obtained in the different body areas and weight categories. Values were log-transformed for the analysis and back-transformed for MT
| Variable | Dog size | Body area | ||||
|---|---|---|---|---|---|---|
| Tibia | Humerus | Neck | T-L | Abdomen | ||
| MT 2 mm probe (N) | Small (1–8 kg) | 4.6 ± 1.6 (1.7–10.50) | 4.3 ± 1.5 (1.62–9.12) | 7.9 ± 1.3 (5.13–11.75) | 5.8 ± 1.6 (1.95–12.02) | 3.4 ± 1.6 (1.74–6.76) |
| Medium (9–22 kg) | 5.6 ± 1.4 (3.63–13.18) | 5.6 ± 1.4 (2.69–12.02) | 9.8 ± 1.3 (5.25–15.49) | 5.9 ± 1.5 (2.29–14.79) | 2.8 ± 1.6 (1.55–5.25) | |
| Large (23–40 kg) | 7.1 ± 1.6 (2.51–18.62) | 7.1 ± 1.4 (3.89–14.79) | 13.5 ± 1.5 (3.39–25.12) | 8.3 ± 1.5 (2.69–20.42) | 4.8 ± 1.7 (1.05–11.22) | |
| MT 4 mm probe (N) | Small (1–8 kg) | 6.5 ± 1.6 (2.45–14.79) | 5.7 ± 1.5 (1.41–15.14) | 9.8 ± 1.4 (3.89–16.98) | 8.1 ± 1.5 (3.09–17.38) | 4.4 ± 1.7 (1.86–8.32) |
| Medium (9–22 kg) | 8.3 ± 1.5 (2.63–14.45) | 7.3 ± 1.3 (4.68–11.75) | 11.9 ± 1.3 (8.13–19.50) | 8 ± 1.4 (3.72–15.140 | 4.3 ± 1.6 (1.86–10.47) | |
| Large (23–40 kg) | 9.5 ± 1.4 (3.8–20.89) | 9.9 ± 1.3 (3.72–16.98) | 16.1 ± 1.3 (7.94–22.91) | 10.3 ± 1.5 (3.31–24.55) | 7.2 ± 1.5 (3.09–15.49) | |
| MT 8 mm probe (N) | Small (1–8 kg) | 9.7 ± 1.3 (5.89–16.22) | 8.9 ± 1.4 (2.69–15.49) | 12.2 ± 1.4 (4.47–19.95) | 12.9 ± 1.4 (6.31–21.88) | 7.2 ± 1.6 (2.63–11.75) |
| Medium (9–22 kg) | 11.1 ± 1.3 (7.24–19.05) | 11.1 ± 1.4 (5.37–18.20) | 15.9 ± 1.3 (9.77–22.91) | 11.4 ± 1.5 (2.75–22.39) | 6 ± 1.5 (2.75–2.75) | |
| Large (23–40 kg) | 13.5 ± 1.4 (6.31–24.55) | 13.8 ± 1.3 (7.41–25.12) | 20.6 ± 1.4 (7.41–34.67) | 15 ± 1.5 (4.47–33.88) | 9.8 ± 1.7 (3.8–29.51) | |
| Heat threshold (°C) | Small (1–8 kg) | 43.0 ± 2.5 (39.10–50.25) | 45.0 ± 3 (40.30–50.87) | 48.2 ± 3.2 (44.10–55.00) | 47.5 ± 3.5 (42.23–55.00) | 44.7 ± 3.3 (40.90–40.90) |
| Medium (9–22 kg) | 43.8 ± 3.1 (39.57–50.20) | 46.6 ± 3.6 (41.40–55.00) | 48.5 ± 3.9 (40.70–55.00) | 47.3 ± 3.5 (40.60–55.00) | 43.9 ± 2.1 (40.70–46.50) | |
| Large (23–40 kg) | 46.4 ± 4 (38.80–55.00) | 49.4 ± 3.7 (39.85–55.00) | 51.9 ± 3.4 (40.00–55.00) | 51.3 ± 3.4 (43.27–55.00) | 46.8 ± 4.5 (37.75–55.00) | |
| TSTUD (gf) | Small (1–8 kg) | (22/32) 68.7 % 79.43 (7.84–597.50) | (22/32) 68.7% 130.80 (8.88–597.50) | (14/32) 43.7% 597.50 (24.05–597.50) | (22/32) 68.7% 164.40 (11.91–597.50) | (8/16) 6.3% 372 (11.91–597.50) |
| Medium (9–22 kg) | (14/18) 77.7% 180 (46.64–597.50) | (6/18) 33.3% 597.50 (72.21–597.50) | (2/18) 11.1% 597.50 (279.1–597.50) | (11/18) 61.1% 311.70 (101.2–597.50) | (5/9) 55.5% 597.50 (71.21–597.50) | |
| Large (23–40 kg) | (24/44) 54.5% 311.70 (6.82–597.50) | (13/44) 29.5% 597.50 (7.55–597.50) | (5/44) 11.3% 597.50 (47.66–597.50) | (23/44) 52.3% 303.10 (7.94–597.50) | (9/22) 40.9% 597.50 (11.66–597.50) | |
| TST50% (gf) | Small (1–8 kg) | (11/16) 68.7 % 300 (180–300) | (6/16) 37.5% 300 (100–300) | (2/16) 12.5% 300 (180–300) | (9/16) 56.2% 300 (180–300) | (2/7) 28.5% 300 (300–300) |
| Medium (9–22 kg) | (10/14) 71.42% 100 (4–300) | (15/18) 83.3% 240 (4–300) | (6/18) 33.3% 300 (180–300) | (17/18) 94.4% 180 (8–300) | (5/7) 71.4% 180 (4–300) | |
| Large (23–40 kg) | (4/4) 100% 37.5 (15–300) | (2/4) 50% 300 (300–300) | (2/4) 50% 300 (300–300) | (4/4) 100% 300 (300–300) | (2/2) 100% 300 (300–300) | |
| Cold (°C) (seconds) | Small (1–8 kg) | (43/123) 35 % 60 (11.41–60) | (18/123) 14.6% 60 (28.17–60) | (18/126) 14.6% 60 (31.40–60) | (15/126) 11.9% 60 (11.97–60) | (4/63) 6.3% 60 (48.33–60) |
| Medium (9–22 kg) | (79/78) 35.2% 58.84 (9.83–60) | (19/75) 25.3% 60 (9.40–60) | (5/78) 6.4% 60 (32.40–60) | (8/78) 10.2% 60 (9.30–60) | (10/39) 25.6% 60 (21.8–60) | |
| Large (23–40 kg) | (71/132) 53.8% 43.25 (18.50–60) | (38/129) 30.2% 60 (41–60) | (38/132) 28.7% 56.36 (6.4–60) | (35/129) 27.1% 60 (9.38–60) | (24/66) 36.6% 54.16 (12.27–60) | |
(gf), gram of force; N, Newton; T-L, thoraco-lumbar area; TSTUD, tactile sensitivity thresholds up–down technique method; TST50%, tactile sensitivity thresholds 50% response technique method.
Table 2Results of linear mixed model and general linear mixed model. Effect of body area, weight category, age category, sex and factors of reliability and performance of the protocol (occasion, feasibility scores and period of testing) on TST, MT, CT, HT
| Fixed effect | TSTUD | TST50% | MT2 mm | M4 mm | MT8 mm | HT | CT |
|---|---|---|---|---|---|---|---|
| BA | 0.783 | 0.001* | <0.001* | <0.001* | <0.001* | <0.001* | <0.001* |
| L/R side | 0.642 | 0.478 | 0.685 | 0.405 | 0.760 | 0.884 | 0.515 |
| Weight category | 0.06 | <0.001* | <0.001* | <0.001* | <0.001* | 0.008* | 0.037* |
| Age category | 0.076 | 0.408 | 0.145 | 0.384 | 0.846 | 0.041* | 0.448 |
| Sex | 0.006* | 0.009* | 0.009* | 0.131 | 0.032* | 0.021* | 0.088 |
| Skin temperature | – | – | – | – | – | 0.457 | 0.082 |
| Room temperature | – | – | – | – | – | 0.365 | 0.087 |
| Feasibility score | 0.004* | 0.060 | 0.557 | 0.144 | 0.852 | 0.08 | 0.221 |
| Occasion (1, 2, 3) | 0.825 | 0.119 | 0.747 | 0.470 | 0.158 | 0.930 | 0.004* |
| Period of testing | 0.573 | – | 0.050* | 0.043* | 0.014* | 0.934 | 0.067 |
BA, body area; CT, cold latency thresholds; HT, heat thresholds; L/R, left/right side; MT, mechanical thresholds; TSTUD, tactile sensitivity thresholds up–down technique method; TST50%, tactile sensitivity thresholds 50% response technique method. * p < 0.005.
Figure 3Median, interquartile range and min–max thresholds of the sensory threshold examination protocol (STEP). The three different weight categories are displayed on the figure. For statistical difference between body areas, see Appendix 3. A, B, C: MT, mechanical threshold with the 2, 4 and 8 mm size probe; N, Newton. D, E: TST, tactile sensitivity threshold; gf, grams of force. F: HT, heat thresholds (°C). G: CT, cold latency (seconds). H: Probability of response to von Frey filaments (TST) on different body areas (%). TSTUD, up–down technique method of testing; TST50%, 50% of response technique method of testing. I: Probability of response to cold stimulus (%).
Influence of explanatory variables
The p values of the different explanatory variables studied are summarized in Table 2. The post hoc comparisons for these effects are reported in Appendix 3 (mean differences and p value for MT and HT; odds ratio and p value for TST and CT). There was a highly significant effect of the BA tested for all stimuli evaluated (p ≤ 0.001). The QST thresholds for the different BA and stimuli are summarized in Fig. 3. Higher thresholds were scored in the neck compared with other areas in all the QST (p ≤ 0.024) (Appendix 3). Left and right sides of each BA showed no significant differences in thresholds in this study (Table 2).
Weight category had a significant effect on all thresholds (p ≤ 0.037) except for pTSTUD. Small dogs had lower MT and HT than medium and large dogs (p ≤ 0.029, Appendix 3). Nevertheless, smaller dogs were less likely to respond to TST50% than larger dogs (p < 0.01). Regarding age, young dogs were more likely to obtain higher HT than adults (p = 0.035); however, adults obtained lower HT than geriatric patients did (p = 0.013). The MT and HT were significantly higher in females (p < 0.05), whereas this effect was not significant for pCT. In contrast, pTST50% was higher in females than in males (p = 0.006 and p = 0.009 for TSTUD and TST50%, respectively).
Test–retest repeatability
There was no inter-occasion difference, except for pCT (Table 2), where percentage of response was significantly higher during the last occasion than the previous two (p < 0.01). Feasibility score only significantly affected pTSTUD (p = 0.004); a higher proportion of responses was obtained with higher feasibility scores (less cooperative dogs). Lower thresholds were obtained for MT on the second period of testing where the operator obtained more experience (p < 0.05) (Appendix 3).
The ICCs showed moderate to substantial test–retest repeatability across occasions (Table 3) except for the TSTUD where the ICC was poor. The two periods of testing showed a significant effect on MT. Therefore, the ICCs of the two periods for MT were calculated. A slight improvement in ICCs was seen (Table 3).
Table 3Intra-class correlation coefficient (ICC) and 95% confidence interval (CI) of the different tests of the sensory threshold examination protocol (STEP) and ICC of mechanical threshold (MT) for the two different periods of testing, where differences in MT were observed in the linear mixed effect model. There is a mild improvement in ICCs between period 1 and period 2 with the three different probes
| TSTUD | TST50% | MT2 | MT4 | MT8 | HT | CT | |
|---|---|---|---|---|---|---|---|
| ICC | 0.001 | 0.71 | 0.72 | 0.69 | 0.68 | 0.58 | 0.51 |
| 95% CI | N/A | 0.1–1 | 0.58–0.86 | 0.52–0.85 | 0.51–0.84 | 0.34–0.86 | 0.22–0.77 |
| Period 1 ICC | N/A | N/A | 0.72 | 0.65 | 0.65 | N/A | N/A |
| Period 2 ICC | N/A | N/A | 0.75 | 0.78 | 0.76 | N/A | N/A |
CT, cold latency thresholds; HT, heat thresholds; TSTUD, tactile sensitivity thresholds up–down technique method; TST50%, tactile sensitivity thresholds 50% response technique method.
Discussion
Canine spontaneous models of chronic pain need a standard procedure for characterization. In addition, investigations of nociception in animals should represent the preliminary step before clinical studies are undertaken to pursue better treatment options in small companion animals (
Bergadano et al. 2006
). This study intended to create and evaluate a STEP to determine a complete QST phenotype in one clinical session. Feasibility, test–retest repeatability, and possible confounding factors (cofactors and covariates) to take into account when applying the STEP were studied.First, consistently with other studies in dogs (
Moore et al., 2013
, Briley et al., 2014
, Harris et al., 2015
), the cofactor that had the largest effect in our study was weight category. Nevertheless, the sample in this study was not large enough to include weight as a continuous explanatory variable, and the diversity of breeds was not representative enough to include this effect in the analysis. Another important factor affecting response is the limb length and the distance between the nociceptor to the brain (Blankenburg et al., 2010
). Practically, thresholds obtained with the STEP should be compared between dogs of the same weight category.Second, different BAs appeared to show very different thresholds, in line with other studies in healthy dogs (
Coleman et al., 2014
, Harris et al., 2015
) and humans (Rolke et al., 2006
). We included different BAs in this protocol so a map of QST thresholds could be evaluated for feasibility and test–retest repeatability and to evaluate if different BAs could show different thresholds as other studies have demonstrated. The choice of BAs in the present study was adapted from previous studies (Coleman et al., 2014
; Harris et al., 2015
) and modified to be performed easily with the tools provided to ensure a good contact and avoid the probe slipping off the tested BA. This may allow different clinicians to use the STEP efficiently and with good results.Neck area scored higher thresholds in all tests of the STEP. There are no other reports of neck thermal or mechanical testing in dogs. It has been suggested that tissues in the more distal aspects of limbs are more highly innervated than more proximal tissues, and nerves have smaller receptive fields (
Coleman et al., 2014
). Contributing factors may also include differences in reaction time related to thickness of the epidermis (Blankenburg et al., 2010
). These findings support the assumption that when testing a patient for sensory abnormalities, thresholds from a specific BA should not be compared with values from a BA of a different location. The lack of differential sensitivity across the left and right sides suggests that the unaffected side of a BA may be an appropriate control for the unilateral affected painful side if this has not been compromised by central sensitization.BAs significantly affected algometer readings in previous studies (
Coleman et al., 2014
; Harris et al., 2015
). MTs for spine and hips reported by Coleman et al., 2014
(mean of approximately 38 and 42 N, respectively) were higher than those for elbows and stifles (mean between 37 and 27 N). It is difficult to compare these results to ours because the testing device differed and large dogs (retrievers) were tested in lateral recumbency—all of which could explain their high MTs (Coleman et al., 2014
). The same finding was reported in studies comparing healthy and osteoarthritic dogs in lateral recumbency (Knazovicky et al., 2016
). The MTs on the tibia with a different device were higher when comparing within the same weight category range of our study (1523 gf being approximately 14.0 versus 9.5 N obtained in our study with the 4 mm tip size). In this case, the tip diameter was 3 mm and the rate of increase of pressure was not indicated. The MTs reported for the different BAs by Harris et al., 2015
with the same device used in our study (i.e. MT of the tibias obtained a mean of 5.6–5.8 N) were not separated by weight. Briley et al., 2014
obtained a mean between 1089 and 1028 gf, which corresponds with approximately 10 N. However, this was on the metatarsal surface, in lateral recumbency and with a different algometer in healthy dogs between 10 and 40 kg, which makes it impossible to compare between studies.There are no other known veterinary studies reporting differences in BA in thermal thresholds in dogs for direct comparison.
Hoffmann et al., 2012
reported a mean HT of 39 °C on the lateral thorax in Beagles weighting 17 kg. Williams et al., 2014
measured the latency of time healthy dogs were able to tolerate standing on a hot infrared light that reached about 59 °C in 30 seconds. Only the hind paw latency was evaluated in this study. Knazovicky et al., 2016
applied a temperature of 45 °C on the tibias and other locations of the pelvic limb and measured latency in large dogs. These areas were not clipped and prevent comparisons between studies.Previously, latency to respond to cold has been evaluated only on a cold plate at 6 °C in the hind paw and the pelvic limb in lateral recumbency in healthy dogs (
Brydges et al., 2012
, Briley et al., 2014
) but not in thoracic limbs, neck or spine. Control dogs reached the cut-off time in most of the cases, as occurred in our study. Knazovicky et al., 2016
reported a mean latency to 0 °C of 52.77 seconds in large dogs in lateral recumbency compared with a median of 43.25 seconds obtained in the tibias in our study. Nevertheless, a standard methodology of testing that allows good test–retest repeatability is necessary to establish a normal range and allow comparison with chronic pain conditions in future studies.Third, age affected the response to testing, as young and geriatric patients showed higher HTs than adults did. Our results are consistent with human studies in which age differences had a large effect in the data (
Rolke et al., 2006
, Blankenburg et al., 2010
). These effects could be related with functional maturation of interneurons in the cortex and dorsal horn when comparing young patients and decrease in innervation density when testing geriatric patients.Fourth, the TST data in this study are in agreement with human studies showing that women tend to be more sensitive to pain than men (
Rolke et al., 2006
). This has been also reported in dogs from the same breed when tested for MT (Coleman et al., 2014
) and may be related with differences in central processing because of genetic and psychological factors (Blankenburg et al., 2010
). However, our results showed the opposite pattern for MT and HT. This could be potentially explained, although not statistically significant, by the higher thresholds obtained by females in the younger group compare to adult group, especially on occasion 3.The von Frey filaments determine a tactile sensory threshold, but not a nociceptive threshold. The TST assesses Aβ fibres (
Hansson et al., 2007
). For the TST, it was impossible to assess the presence of mechanical allodynia since it was not present in the sample of healthy dogs tested, and the % of response to any the von Frey filaments was very variable (Table 1). A similar pattern was observed with CT, where latency at which the cold stimulus (0 °C) may become nociceptive (assessment of Aδ and C fibres) could not be established owing to the lack of response to cold in some dogs/BAs. The upper limits for HT and CT are actually the upper possible safety limits; therefore, a true upper range could not be obtained in this case (censored data). These problems have also been reported in healthy human volunteers (Rolke et al., 2006
). Briley et al., 2014
studied the feasibility of the same device used in our study, demonstrating similarly to our finding large variability of response to 0 °C during the same cut-off time, with healthy dogs. Dogs with osteoarthritis and spinal cord injury showed lower latencies to 0 °C compared to healthy dogs (Knazovicky et al., 2016
, Gorney et al., 2016
). However, further studies in dogs with different pain modalities are needed to elucidate whether this device could be used as a tool to detect allodynia or hyperalgesia, as it seems that 0 °C did not trigger a nociceptive response within 60 seconds in all healthy dogs.Two methods to evaluate TST were compared in this study. The TST50% has been used previously in dogs with cranial cruciate ligament rupture (
Brydges et al., 2012
), showing good results in identifying individuals with central sensitization. These authors reported a mean of 900 mN mm−2 in control dogs between the second and the third digits of the pelvic limb, which corresponds approximately with 300 gf (similar to our findings). It seemed that, although still variable, a higher proportion of healthy dogs responded below the cut-off with the TST50%. The present study showed that the TST50% technique was more repeatable, with less variability between subjects and behaved similarly to other tests regarding factors influencing results such as weight category, gender and BAs when compared with the TSTUD. In contrast, the TSTUD did not have a good utility in healthy dogs.For mechanical thresholds, methods of testing need standardization as wider tip diameters have been associated with higher thresholds and a large data range or between-individual variability (higher SD) in previous studies (
Harris et al., 2015
, Taylor et al., 2016a
). Our results show similar ICCs for the different probe sizes with only slightly higher repeatability using the 2 mm probe as previously reported (Harris et al., 2015
). However, other studies used different methods of assessment of test–retest repeatability (Harris et al., 2015
, Taylor et al., 2016a
).In veterinary medicine, the reliability of QST has been assessed with different methods to evaluate variation in QST thresholds over time (
Williams et al., 2014
, Brydges et al., 2012
, Moore et al., 2013
, Briley et al., 2014
, Gorney et al., 2016
, Song et al., 2016
). It has been suggested that the most appropriate method to report test–retest repeatability when exploring QST protocols (Moloney et al., 2012
) is the ICC in conjunction with a measure of precision [i.e., 95% confidence interval (CI)]. However, this method has its limitations, especially if the 95% CI is large as occurred for TST and CT in this study. When the variability between individuals is very large, it can also provide a falsely good ICC, and should be interpreted with caution (Lee et al., 2012
).Chong and Cros, 2004
defined QST evaluation as a subjective psychophysical test, where the consistency of the data relies on environmental factors, methodological factors and the attention and cooperation of the individual being tested. To help with this possible bias in our study, a feasibility score adapted from a previous study assessing mechanical and thermal thresholds in dogs in lateral recumbency (Briley et al., 2014
) was used to evaluate cooperation of dogs and reaction to the stimuli. Feasibility score only affected pTSTUD; thus, overall we found good cooperation >50% of the time, mild sensitivity to being touched and mild variation in reaction to stimuli, sufficient to ensure a good feasibility and repeatability of the STEP. A higher proportion of dogs responded in the third testing occasion for CT, probably trying to avoid an uncomfortable sensation learned from previous tests. Other studies evaluating mechanical testing with other devices also showed a learning effect (Coleman et al., 2014
).An effect of the operator's experience was also evident for MT. During the second period of testing, not only were MTs lower but ICCs were also slightly better compared with first period, and thorough operator training is advised before clinical use. Standardization of instructions to subjects, training of technicians, machine calibration, stimulus characteristics, and testing algorithms are all essential for accurate and reproducible QST (
Chong and Cros, 2004
).Protocols involving QST evaluation in humans include verbal communication of detection thresholds. In veterinary patients, this approach cannot be used, and instead reliance must be placed on observable behavioural indicators. In the case of animals with peripheral and central sensitization, where somatosensory function evaluated by QST encompasses the presence of allodynia or hyperalgesia as well as pain, it is not possible to reliably distinguish between thresholds of sensation and nociception. Consequently, some authors view QST as a semiobjective assessment (
Gorney et al., 2016
). Nevertheless, QST can provide valuable clinical information regarding the impacts on patients (Brown, 2012
).Limitations of the study include the small number of dogs tested. Further data may be required to obtain reliable reference values. In future studies, dogs with inability to stand may not be suitable for the current protocol. Position (sitting, laying in lateral recumbency) has been tested in other studies (
Harris et al., 2015
; Knazovicky et al., 2016
, Gorney et al., 2016
) and could be a possibility for these patients. Fatigue from remaining standing was accounted for, and short periods of resting were allowed between tests. Clipping may not be possible in some patients with severe allodynia, and the full battery of tests may not be possible to perform in that particular BA: instead, other diagnostic tools could compliment the assessment, including history, imaging tests, chronic pain questionnaires and behavioural response when approaching the area.In conclusion, the sensory testing examination protocol showed substantial to moderate test–retest repeatability for HT and MT in healthy dogs. The STEP was feasible, safe and well tolerated. Cold and tactile sensitivity thresholds showed poor consistency in response to the stimuli, and ICCs showed heterogeneity across these data. Further work in dogs with central sensitization is needed to assess the usefulness and test–retest repeatability of the STEP in practice. Testing only the specific BA of interest could be envisaged to shorten the duration of the protocol when phenotyping different pain conditions. Since weight category was the most significant explanatory variable, nociceptive thresholds for the STEP were displayed based on this covariate and in the future should only be compared within weight class. Further studies in dogs with painful conditions should evaluate the utility of each test in detecting sensory abnormalities in dogs.
Acknowledgements
The authors thank Heather Williams, Jo Murrell and Nicolas Granger from the University of Bristol for their collaborative support in this area of research. The authors are very grateful to Transpharmation Ltd. for supporting this work. Many thanks to Mrs Carol Hoy Ncert A&CC, VTS (Anesthesia/Analgesia), RVN for the figure provided. Thank you to all the dog owners for participating in this study.
Authors' contributions
SSM: design, data management, data interpretation, statistical analysis and preparation of manuscript; YC: data interpretation, statistical analysis and preparation of manuscript; SA: data interpretation, statistical analysis and preparation of manuscript; AF: data interpretation and preparation of manuscript; HAV: data interpretation and preparation of manuscript; LP: design, data management, data interpretation, statistical analysis and preparation of manuscript.
Conflict of interest statement
Authors declare no conflict of interest.
Appendix 1. Feasibility scores. Adapted from (Briley et al. 2014).
Tabled
1
| Feasibility score | Description |
|---|---|
| 0 – No problem | Minimum restraint needed; excellent cooperation; clear reaction to stimuli |
| 1 – Mild difficulty | Mild restraint needed; good cooperation; clear reaction to stimuli |
| 2 – Moderate difficulty | Moderate restraint needed; good cooperation >50% of the time; mild sensitivity to being touched; mild variation in reaction to stimuli |
| 3 – Significant difficulty | Significant restraint needed and resisted sternal position; good cooperation <25% of the time; moderate sensitivity to being touched; moderate variation in reaction to stimuli |
| 4 – Extreme difficulty | Constant restraint required; not cooperative; unclear reaction to stimuli, not confident in data collected |
| 5 – Impossible | Could not collect data due to the dog's disposition and/or lack of confidence in the reactions seen being due to the stimulus |
Appendix 2. Body weight and age categories of the sample of dogs.
Tabled
1
| Category | Classification | Dogs | |
|---|---|---|---|
| n | % | ||
| Age (Years) | Young (0.3–3) | 9 | 36 |
| Adult (4–6) | 9 | 36 | |
| Senior (> 6) | 7 | 28 | |
| Weight (kg) | Small (1–8) | 10 | 40 |
| Medium (9–22) | 6 | 24 | |
| Large (23–40) | 9 | 36 | |
n, number of dogs.
Appendix 3. Post Hoc comparisons, odds ratio (OR) and estimated mean differences comparing body areas, weight category, age category, sex, feasibility score, occasion tested and period of testing. Main differences for mechanical thresholds (MT) 2, MT4 and MT8 are displayed as back log transformed.
Tabled
1
| Pairwise comparison | pTSTUD | pTST50% | MT2 (N) | MT4 (N) | MT8 (N) | HT (°C) | pCT | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OR | P-value | OR | P-value | Mean difference | P-value | Mean difference | P-value | Mean difference | P-value | Mean difference | P-value | OR | P-value | |
| BA | ||||||||||||||
| Tibia–Humerus | – | – | 0.3 | 0.105 | 1.0 | 0.429 | 1.1 | 0.140 | 1.01 | 0.697 | −2.7 | <0.001* | 0.7 | <0.001* |
| Tibia–Neck | – | – | 0.0 | <0.001* | −1.7 | <0.001* | −1.6 | <0.001* | −1.4 | <0.001* | −5.4 | <0.001* | 0.7 | <0.001* |
| Tibia–T-L | – | – | 1.1 | 0.936 | −1.1 | 0.003* | −1.1 | 0.023* | −0.3 | 0.001* | −4.5 | <0.001* | 0.7 | <0.001* |
| Tibia–Abdomen | – | – | 0.4 | 0.304 | 1.6 | <0.001* | 1.5 | <0.001* | 1.5 | <0.001* | −0.9 | 0.023* | 0.7 | <0.001* |
| Humerus–neck | – | – | 0.1 | 0.008* | −1.8 | <0.001* | −1.6 | <0.001* | −1.4 | <0.001* | −2.7 | <0.001* | 0.9 | 0.028* |
| Humerus–T-L | – | – | 3.7 | 0.083 | −1.1 | <0.001* | −1.2 | <0.001* | −1.1 | <0.001* | −1.7 | <0.001* | 0.9 | 0.032* |
| Humerus–Abdomen | – | – | 1.3 | 0.768 | 1.5 | <0.001* | 1.4 | <0.001* | 1.5 | <0.001* | 1.7 | 0.001* | 1.0 | 0.642 |
| Neck–T-L | – | – | 24.5 | <0.001* | 1.5 | <0.001* | 1.4 | <0.001* | 1.2 | <0.001* | 0.9 | 0.014* | 1.0 | 0.850 |
| Neck–Abdomen | – | – | 9.0 | 0.024* | 2.8 | <0.001* | 2.4 | <0.001* | 2.1 | <0.001* | 4.4 | <0.001* | 1.0 | 0.084 |
| T-L–Abdomen | – | – | 0.3 | 0.269 | 1.8 | <0.001* | 1.7 | <0.001* | 1.7 | <0.001* | 3.5 | <0.001* | 1.1 | 0.095 |
| Weight category | ||||||||||||||
| Small–Medium | – | – | 0.03 | <0.001* | −1.4 | 0.008* | −1.3 | 0.029* | −1.4 | 0.012* | −1.1 | 0.316 | 1.2 | 0.063 |
| Small–Large | – | – | 0.0000008 | 0.001* | −1.9 | <0.001* | −1.6 | <0.001* | −1.8 | <0.001* | −0.3 | 0.001* | 1.5 | 0.010* |
| Medium–Large | – | – | 0.00002 | 0.006* | −1.3 | 0.010* | −1.3 | 0.005* | −1.3 | 0.014* | −2.6 | 0.020* | 1.1 | 0.485 |
| Age category | ||||||||||||||
| Young–Adults | – | – | – | – | – | – | – | – | – | – | 2.2 | 0.035* | – | – |
| Young–Senior | – | – | – | – | – | – | – | – | – | – | −0.6 | 0.562 | – | – |
| Adults–Senior | – | – | – | – | – | – | – | – | – | – | −2.9 | 0.013* | – | – |
| Sex | ||||||||||||||
| Female–Male | 0.5 | 0.006* | 0.04 | 0.009* | 1.3 | 0.009* | – | – | 1.2 | 0.023* | 2.5 | 0.021* | – | – |
| Feasibility score | ||||||||||||||
| 0–1 | 0.3 | 0.001* | – | – | – | – | – | – | – | – | −1.2 | 0.296 | – | – |
| 0–2 | 0.4 | 0.028* | – | – | – | – | – | – | – | – | 1.5 | 0.153 | – | – |
| 1–2 | 1.3 | 0.214 | – | – | – | – | – | – | – | – | 2.7 | 0.013* | – | – |
| Occasion | ||||||||||||||
| 1–2 | – | – | – | – | – | – | – | – | – | – | – | – | 1.1 | 0.223 |
| 1–3 | – | – | – | – | – | – | – | – | – | – | – | – | 0.9 | 0.006* |
| 2–3 | – | – | – | – | – | – | – | – | – | – | – | – | 0.8 | 0.003* |
| Period of testing | ||||||||||||||
| 1–2 | – | – | – | – | −1.2 | 0.050* | −1.2 | 0.043* | −9.1 | 0.014* | – | – | – | – |
BA, body area; pTSTUD, response to tactile sensitivity up-down technique method and pTST50% with 50% response technique method (any of the von Frey filaments); MT, mechanical thresholds; HT: heat thresholds; pCT: response to 0° C before 60 seconds; N: newton; T-L: thoraco-lumbar area; (–), no significant difference for covariate/cofactor on this test; *,P < 0.05.
References
- Test–retest agreement and reliability of quantitative sensory testing 1 year after breast cancer surgery.Clin J Pain. 2014; 31: 393-403
- Value of quantitative sensory testing in neurological and pain disorders: NeuPSIG consensus.Pain. 2013; 154: 1807-1819
- Neuropathic pain: precise sensory profiling improves treatment and calls for back-translation.Pain. 2014; 155: 2215-2217
- Quantitative assessment of nociceptive processes in conscious dogs by use of the nociceptive withdrawal reflex.Am J Vet Res. 2006; 67: 882-889
- Reference values for quantitative sensory testing in children and adolescents: developmental and gender differences of somatosensory perception.Pain. 2010; 149: 76-88
- Feasibility and repeatability of cold and mechanical quantitative sensory testing in normal dogs.Vet J. 2014; 199: 245-250
- Ability of the canine brief pain inventory to detect response to treatment in dogs with osteoarthritis.J Am Vet Med Assoc. 2008; 233: 1278-1283
- Quantitative sensory testing: a stimulating look at chronic pain.Vet J. 2012; 193: 315-316
- Clinical assessments of increased sensory sensitivity in dogs with cranial cruciate ligament rupture.Vet J. 2012; 193: 545-550
- Quantitative assessment of tactile allodynia in the rat paw.J Neurosci Methods. 1994; 53: 55-63
- Technology literature review: quantitative sensory testing.Muscle Nerve. 2004; 29: 734-747
- Learning confounds algometric assessment of mechanical thresholds in normal dogs.Vet Surg. 2014; 43: 361-367
- A thermal threshold testing device for evaluation of analgesics in cats.Res Vet Sci. 2002; 72: 205-210
- Patient phenotyping in clinical trials of chronic pain treatments: IMMPACT recommendations.Pain. 2016; 157: 1851-1871
- Neuropathic pain: an updated grading system for research and clinical practice.Pain. 2016; 157: 1599-1606
- Mechanical and thermal sensory testing in normal chondrodystrophoid dogs and dogs with spinal cord injury caused by thoracolumbar intervertebral disc herniations.J Vet Intern Med. 2016; 30: 627-635https://doi.org/10.1111/jvim.13913
- Quantitative assessment of neuropathic pain.Curr Pain Headache Rep. 2001; 5: 107-113
- Test order of quantitative sensory testing facilitates mechanical hyperalgesia in healthy volunteers.J Pain. 2012; 13: 73-80
- Neuropathic pain: clinical characteristics and diagnostic workup.Eur J Pain. 2002; 6: 47-50
- Usefulness and limitations of quantitative sensory testing: clinical and research application in neuropathic pain states.Pain. 2007; 129: 256-259
- Influence of experimental protocol on response rate and repeatability of mechanical threshold testing in dogs.Vet J. 2015; 204: 82-87
- Contact heat thermal threshold testing in beagle dogs: baseline reproducibility and the effect of acepromazine, levomethadone and fenpipramide.BMC Vet Res. 2012; 8: 206
- Sedative and analgesic effects of buprenorphine, combined with either acepromazine or dexmedetomidine, for premedication prior to elective surgery in cats and dogs.Vet Anaesth Analg. 2013; 40: 297-307
- Widespread somatosensory sensitivity in naturally occurring canine model of osteoarthritis.Pain. 2016; 157: 1325-1332https://doi.org/10.1097/j.pain.0000000000000521
- The measurement of observer agreement for categorical data.Biometrics. 1977; 33: 159-174
- Getting a sense of sensations.Vet J. 2013; 197: 115-117
- Pitfalls and important issues in testing reliability using intraclass correlation coefficients in orthopaedic research.Clin Orthop Surg. 2012; 4: 149-155
- Neuropathic pain in dogs and cats: if only they could tell us if they hurt.Vet Clin Small Anim. 2008; 38: 1365-1414
- Reliability of thermal quantitative sensory testing: a systematic review.J Rehabil Res Dev. 2012; 49: 191-207
- The use of an electronic von Frey device for evaluation of sensory threshold in neurologically normal dogs and those with acute spinal cord injury.Vet J. 2013; 197: 216-219
- Phenotyping neuropathic pain patients: implications for individual therapy and clinical trials.Curr Opin Support Palliat Care. 2014; 8: 124-129
- Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values.Pain. 2006; 123: 231-243
- von Frey anesthesiometry to assess sensory impairment after acute spinal cord injury caused by thoracolumbar intervertebral disc extrusion in dogs.Vet J. 2016; 209: 144-149
- Mechanical nociceptive thresholds using four probe configurations in horses.Vet Anaesth Analg. 2016; 43: 99-108https://doi.org/10.1111/vaa.12274
- Relationship between mechanical thresholds and limb use in dogs with coxofemoral joint OA-associated pain and the modulating effects of pain alleviation from total hip replacement on mechanical thresholds.Vet Surg. 2014; 43: 542-548
- Applying linear mixed models to estimate reliability in clinical trial data with repeated measurements.Control Clin Trials. 2004; 25: 13-30
- Quantitative sensory testing and mapping: a review of nonautomated quantitative methods for examination of the patient with neuropathic pain.Clin J Pain. 2009; 25: 632-640
- An exploratory study into objective and reported characteristics of neuropathic pain in women with chronic pelvic pain.PLoS One. 2016; 11: e0151950
- Feasibility and repeatability of thermal quantitative sensory testing in normal dogs and dogs with hind limb osteoarthritis-associated pain.Vet J. 2014; 199: 63-67
Article Info
Publication History
Published online: February 23, 2017
Accepted:
September 2,
2016
Received:
February 22,
2016
Identification
Copyright
© 2017 Association of Veterinary Anaesthetists and American College of Veterinary Anesthesia and Analgesia. Published by Elsevier Ltd. All rights reserved.
30070-3&id=gr1.jpg){kind=link}
30070-3&id=gr2.jpg){kind=link}
30070-3&id=gr3.jpg){kind=link}