Advertisement

The Canine Postamputation Pain (CAMPPAIN) initiative: a retrospective study and development of a diagnostic scale

Published:August 12, 2021DOI:https://doi.org/10.1016/j.vaa.2021.07.003

      Abstract

      Objective

      To develop a scale to diagnose and assess the severity of postamputation pain (PAP) in dogs.

      Study design

      Single-center retrospective study.

      Animals

      A total of 66 dogs that underwent thoracic or pelvic limb amputation and 139 dogs that underwent tibial plateau leveling osteotomy (TPLO) at a veterinary teaching hospital.

      Methods

      An online survey regarding postoperative behavioral changes was sent to owners. Categorical, multiple-choice responses were entered into a univariable logistic regression model and tested for association with amputation using the Wald test. If p < 0.2, variables were forwarded to a multivariable logistic regression model for manual build. Model simplicity and predictive ability were optimized using the area under the receiver operating curve (AUROC) characteristic, and model calibration was assessed using the Hosmer–Lemeshow test. The selected model was converted to an integer scale (0–10), the Canine Postamputation Pain (CAMPPAIN) scale. Univariable logistic regression related each dog’s calculated score to the probability of PAP.

      Results

      Multivariable logistic regression identified four independent predictors of PAP (p < 0.05): 1) restlessness or difficulty sleeping, 2) episodes of panic or anxiety, 3) sudden vocalization, and 4) compulsive grooming of the residual limb. Score AUROC was 0.70 (95% confidence interval = 0.63–0.78) with good calibration (Hosmer–Lemeshow statistic p = 0.82). A score of 2 corresponded to a risk probability of 0.5. Taking a score ≥ 2 to indicate PAP, score specificity and sensitivity were 92.1% and 36.4%, respectively. When this score was used to diagnose PAP, prevalence was 36.4% (24/66) and 7.9% (11/139) in the amputation and TPLO groups, respectively.

      Conclusions and clinical relevance

      Postamputation pain is characterized by specific postoperative behaviors and appears to affect approximately one-third of canine amputees. The CAMPPAIN scale generated from these data could facilitate diagnosis, treatment and further study of PAP but requires external validation.

      Keywords

      Introduction

      Postamputation pain (PAP) in humans was first documented in the medical literature by the military surgeon Ambroise Paré almost 500 years ago (
      • Paré A.
      La maniere de traicter les playes faictes tant par hacquebutes: que par fleches: & les accidentz d'icelles, comme fractures & caries des os, grangrene & mortification; avec les pourtraictz des instrumentz necessaires pour leur curation. Et La methode de curer les combustions principalement faictes par la pouldre a canon.
      ). Today, PAP is well described in human patients, due largely to the many amputations in conflict zones around the world; although traffic and other accidents, vascular disease, infection and cancer may also necessitate amputation in people (
      • Ziegler-Graham K.
      • MacKenzie E.J.
      • Ephraim P.L.
      • et al.
      Estimating the prevalence of limb loss in the United States: 2005 to 2050.
      ;
      • Alviar M.J.M.
      • Hale T.
      • Dungca M.
      Pharmacologic interventions for treating phantom limb pain.
      ).
      The Postamputation Pain Algorithm (PAPA) developed for humans by the Durham Pain Investigations Group divides PAP into phantom limb pain (PLP, pain perceived in a missing limb or other body part) and residual limb pain (RLP, pain perceived in the residual limb) (
      • Clarke C.
      • Lindsay D.R.
      • Pyati S.
      • Buchheit T.
      Residual limb pain is not a diagnosis: a proposed algorithm to classify postamputation pain.
      ). A review of PAP is beyond the scope of this article, and readers are referred to other sources on this topic (
      • Clarke C.
      • Lindsay D.R.
      • Pyati S.
      • Buchheit T.
      Residual limb pain is not a diagnosis: a proposed algorithm to classify postamputation pain.
      ;
      • Collins K.L.
      • Russell H.G.
      • Schumacher P.J.
      • et al.
      A review of current theories and treatments for phantom limb pain.
      ;
      • Kuffler D.P.
      Origins of phantom limb pain.
      ). PLP is an incompletely understood, complex pain state involving the central nervous system (CNS) and peripheral nervous system; it has a neurophysiological basis and is not solely psychogenic as once thought (
      • Machin P.
      • Williams A.C.
      Stiff upper lip: coping strategies of World War II veterans with phantom limb pain.
      ;
      • Collins K.L.
      • Russell H.G.
      • Schumacher P.J.
      • et al.
      A review of current theories and treatments for phantom limb pain.
      ;
      • Kuffler D.P.
      Origins of phantom limb pain.
      ). By contrast, RLP can have both nociceptive (i.e., caused by nociceptor stimulation in the residual limb) and neuropathic (e.g., caused by neuroma) components (
      • Clarke C.
      • Lindsay D.R.
      • Pyati S.
      • Buchheit T.
      Residual limb pain is not a diagnosis: a proposed algorithm to classify postamputation pain.
      ). Both PLP and RLP are distinct from the postoperative nociceptive pain caused by surgical tissue handling in all patients (
      • Hanley M.A.
      • Jensen M.P.
      • Smith D.G.
      • et al.
      Preamputation pain and acute pain predict chronic pain after lower extremity amputation.
      ). In humans, PAP can trigger depression and anxiety and profoundly affect virtually every aspect of daily living, substantially decreasing quality of life (
      • Padovani M.T.
      • Martins M.R.I.
      • Venâncio A.
      • Forni J.E.N.
      Anxiety, depression and quality of life in individuals with phantom limb pain.
      ).
      In general, amputation in the dog produces positive outcomes. Most dogs are fully functional shortly after surgery, and most owners are satisfied with their decision to amputate (
      • Carberry C.A.
      • Harvey H.J.
      Owner satisfaction with limb amputation in dogs and cats.
      ;
      • Kirpensteijn J.
      • van den Bos R.
      • Endenburg N.
      Adaptation of dogs to the amputation of a limb and their owners' satisfaction with the procedure.
      ;
      • Dickerson V.M.
      • Coleman K.D.
      • Ogawa M.
      • et al.
      Outcomes of dogs undergoing limb amputation, owner satisfaction with limb amputation procedures, and owner perceptions regarding postsurgical adaptation: 64 cases (2005-2012).
      ;
      • Galindo-Zamora V.
      • von Babo V.
      • Eberle N.
      • et al.
      Kinetic, kinematic, magnetic resonance and owner evaluation of dogs before and after the amputation of a hind limb.
      ). However, behavioral changes such as aggression, anxiety, compulsive licking or biting of the residual limb and vocalization have been reported in dogs after amputation and may reflect the presence of PAP (
      • Gross T.L.
      • Carr S.H.
      Amputation neuroma of docked tails in dogs.
      ;
      • Kirpensteijn J.
      • van den Bos R.
      • Endenburg N.
      Adaptation of dogs to the amputation of a limb and their owners' satisfaction with the procedure.
      ;
      • Menchetti M.
      • Gandini G.
      • Gallucci A.
      • et al.
      Approaching phantom complex after limb amputation in the canine species.
      ;
      • Ferreira M.G.
      • Antunes A.V.
      • Pascoli A.L.
      • et al.
      Neuropathic pain and prognosis of a dog submitted to limb amputation after diagnosis of soft tissue sarcoma - case report.
      ). In the authors’ clinical experience, some canine amputees display behaviors consistent with descriptions of neuropathic pain in human amputees.
      Although PAP has been hypothesized to be incidental in canine amputee populations, there have been limited attempts to date to formally differentiate it from the nociceptive pain expected in any dog after surgery [e.g., a major orthopedic surgery such as tibial plateau leveling osteotomy (TPLO)]. The objective of this study was to use owner responses to a survey of postoperative behavioral changes in dogs that had undergone amputation or TPLO to develop a scale that could be used clinically to diagnose and assess the severity of PAP in dogs. The hypothesis was that PAP could be characterized by a constellation of postoperative behaviors unique to amputees. Based on the authors’ clinical experience, it was anticipated that one of these behaviors would be spontaneous, paroxysmal vocalization.

      Materials and methods

      This study was designed in accordance with the guidelines in the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement (
      • von Elm E.
      • Altman D.G.
      • Egger M.
      • et al.
      The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies.
      ).

       Survey design

      An online survey was developed using Qualtrics software (Drive Provo, UT, USA) by a veterinary anesthesiologist (JB), surgeon (DL) and student (KR), with input from a behaviorist (Appendix SA). It consisted primarily of multiple-choice questions based on clinical signs described in canine amputees in other publications and the investigators’ clinical experience with behavioral changes in canine amputees (
      • Kirpensteijn J.
      • van den Bos R.
      • Endenburg N.
      Adaptation of dogs to the amputation of a limb and their owners' satisfaction with the procedure.
      ;
      • Menchetti M.
      • Gandini G.
      • Gallucci A.
      • et al.
      Approaching phantom complex after limb amputation in the canine species.
      ;
      • Ferreira M.G.
      • Antunes A.V.
      • Pascoli A.L.
      • et al.
      Neuropathic pain and prognosis of a dog submitted to limb amputation after diagnosis of soft tissue sarcoma - case report.
      ). A brief description of the study objective (identification of signs of pain after amputation or TPLO) was described within the e-mail. ‘Skip logic’ was used so that questions pertaining to a previous response only appeared if relevant to the previous answer. All questions (the number of which varied related to the use of skip logic) were mandatory before a respondent could continue. After some questions, a box was provided for owners to expand upon or clarify their answers; these boxes were optional.

       Identification of amputation group

      Dogs that underwent amputation in the Cornell University Hospital for Animals (CUHA) from June 2007 through March 2019 were identified via query for ‘amputation’ in the CUHA electronic medical record system [Universal Veterinary Information System (UVIS); Ross Group Inc., OH, USA] or a procedure code corresponding to amputation in a case log spreadsheet (Microsoft Excel 2016; Microsoft, WA, USA) maintained by the surgery service. Dogs of any age, sex or breed that underwent amputation for any reason were included. After all survey responses had been collected, exclusion criteria were applied prior to data analysis to increase uniformity in the amputation group (Fig. 1).
      Figure 1
      Figure 1Flow diagram showing the number of cases (dogs undergoing amputation) potentially eligible for the study (n = 333), the number of cases after each exclusion criterion was applied and the final number of cases for analysis (n = 66).

       Identification of TPLO group

      Dogs that underwent TPLO in the CUHA from March 2007 through June 2015 were identified using a list from another study (
      • Lopez D.J.
      • VanDeventer G.M.
      • Krotscheck U.
      • et al.
      Retrospective study of factors associated with surgical site infection in dogs following tibial plateau leveling osteotomy.
      ). Any age, sex or breed was included.
      Dogs that underwent TPLO were selected as the comparison group because: 1) TPLO was the most common orthopedic surgery performed at the authors’ institution; 2) amputation and TPLO are commonly performed in large-breed dogs, minimizing size and breed differences; and 3) TPLO is expected to cause pain that is quantitatively similar to PAP (i.e., severe pain) but might be qualitatively different, optimizing the likelihood of identifying behaviors specific to PAP. In other words, whereas both amputation and TPLO result in severe nociceptive pain (pain that arises from damage to non-neural tissue and resulting from stimulation of nociceptors), TPLO, like total knee arthroplasty in humans, might be less likely to produce neuropathic pain (pain from a lesion or disease of the somatosensory nervous system) (
      • Kooijman C.M.
      • Dijkstra P.U.
      • Geertzen J.H.B.
      • et al.
      Phantom pain and phantom sensations in upper limb amputees: an epidemiological study.
      ;
      Task Force on Taxonomy
      IASP Terminology. International Association for the Study of Pain (IASP).
      ;
      • Fitzsimmons M.
      • Carr E.
      • Woodhouse L.
      • Bostick G.P.
      Development and persistence of suspected neuropathic pain after total knee arthroplasty in individuals with osteoarthritis.
      ).
      The study design sought two dogs that had undergone TPLO for each dog that had undergone amputation. These could not be matched based on signalment or date of surgery because of the limited number of owner responses and problems with the medical record query function.

       Patient management

      For both groups, a board-certified anesthesiologist approved the anesthetic/analgesic plan, which was carried out by an anesthesiology resident, licensed veterinary technician and/or veterinary student under the anesthesiologist’s supervision. A board-certified surgeon (or resident under supervision) performed amputation or TPLO according to standard techniques (
      • Kowaleski M.
      • Boudrieau R.
      • Pozzi A.
      Stifle joint.
      ;
      • Seguin B.
      • Weigel J.
      Amputations.
      ).

       Survey administration

      Exemption from institutional review board approval was obtained. The survey was sent electronically to owners using the e-mail addresses provided in the medical record. If the e-mail address was not provided, or an owner did not complete the survey after 2 weeks, these owners were called using the telephone number provided in the medical record and asked to participate (and provide their e-mail address if needed). Reminder e-mails were sent periodically to owners who had not yet completed the survey. Owners were called twice at 48 hour intervals before being removed from the calling list.

       Data collection

      Completed survey responses were imported into a spreadsheet (Microsoft Excel 2016). Additional data about each dog was retrieved from CUHA electronic medical records systems (UVIS; ezyVet, TX, USA and/or Perceptive Content; Hyland Software Inc., OH, USA), including signalment and body mass on the day of surgery, reason for amputation, details about the surgical procedure and any complications, and date of death (if applicable).

       Statistical analysis

      All analyses were performed using Stata Version 15 (StataCorp, TX, USA). Normality was assessed using the Shapiro–Wilk test. Continuous data were expressed as mean ± standard deviation if normally distributed and median (interquartile range) if not normally distributed. Between-group differences were assessed using the Student t test if data were normally distributed, and the Mann–Whitney test if data were not normally distributed. Categorical data were presented as proportions. Categorical differences were assessed using the Fisher’s exact test where cell number was ≤ 5 and the chi-square test where > 5.
      If an owner responded ‘unknown or can’t remember’, this response was recorded as equivalent to ‘never’ for the purposes of analysis. Answers to multiple-choice questions about postoperative behaviors (questions 7–32) were converted to integers and modelled as categorical data. These values were entered into a univariable logistic regression model and tested for association with amputation using the Wald test. Behavioral variables were included in a multivariable model if they achieved significance (p < 0.2) in the univariable analysis. A multivariable logistic regression model was then constructed, optimizing model simplicity and predictive ability using the area under the receiver operating curve (AUROC) characteristic. Model calibration was assessed using the Hosmer–Lemeshow test.
      The selected logistic regression model was then converted to an integer scale, the Canine Postamputation Pain (CAMPPAIN) scale, to distinguish PAP from post-TPLO pain using standard methods as follows (
      • Hosmer D.W.
      • Lemeshow S.
      Applied Logistic Regression.
      ;
      • Hayes G.
      • Mathews K.
      • Doig G.
      • et al.
      The acute patient physiologic and laboratory evaluation (APPLE) score: a severity of illness stratification system for hospitalized dogs.
      ). The desired maximum score was selected arbitrarily as 10. The lowest yield categories for each behavioral variable were set as the referent (score = 0) categories. The coefficients assigned to the highest yield category for each behavioral variable in the model were then summed. The desired maximal score for the model was divided by this value to derive a constant. All coefficients were then multiplied by this value and rounded to the nearest 0.5 to obtain the integer scores for each behavioral variable. Univariable logistic regression was then used to relate the calculated score for each dog to the probability of the presence of PAP. The score AUROC and calibration was assessed within this case population. Because of the relatively small number of dogs in each group and the large variability in the perioperative anesthetic/analgesic protocols, associations between drugs or locoregional techniques and postoperative behavioral changes were not investigated.

      Results

       Population characteristics

      Medical record query identified 333 dogs that had undergone amputation and 612 dogs that had undergone TPLO over the specified time periods. Survey response rates were similar at 27.0% and 22.7%, respectively (p > 0.05). After applying exclusion criteria (Fig. 1), 66 dogs with complete surveys remained in the amputation group, which consisted of 37 thoracic limb amputations [scapulectomy, 35/37 (94.6%); distal radius, 1/37 (2.7%); or radiocarpal joint, 1/37 (2.7%)] and 29 pelvic limb amputations [coxofemoral disarticulation, 22/29 (75.9%); acetabulectomy, 1/29 (3.4%); mid-femur, 5/29 (17.2%); or mid-tibia, 1/29 (3.4%)]. The reasons for amputation were bony neoplasia (n = 40), trauma (n = 14), soft tissue neoplasia (n = 7), failure of an orthopedic procedure (n = 4) and vascular ischemia (n = 1). A surgical complication (cellulitis) developed in one dog postoperatively that resolved with antibiotic treatment. This dog displayed signs of pain including restlessness or difficulty sleeping (at least daily), episodes of panic or anxiety (at least weekly) and spontaneous, paroxysmal vocalization (at least daily) that resolved 1–6 months after amputation. Complete surveys of 139 dogs remained in the TPLO group, of which 92/139 (66.2%) underwent unilateral TPLO, 40/139 (28.8%) underwent bilateral TPLO on different days, and 7/139 (5.0%) underwent bilateral TPLO on the same day. Surgical complications (infection) developed in 13/139 (9.4%) dogs.
      The amputation group was older (7.4 ± 3.4 versus 5.7 ± 2.7 years, p < 0.01) and weighed less [32 (22.6–39.5) versus 35.3 (28–41.1) kg, p = 0.04]. There was no difference in sex distribution between groups (p = 0.14). The most common breeds in the amputation and TPLO groups were mixed breed [12/66 (18.2%) and 39/139 (28.1%), respectively] and Labrador Retriever [10/66 (15.2%) and 44/139 (31.7%), respectively]. The proportion of dogs alive at the time of survey completion was greater in the TPLO than in the amputation group (71.2% versus 56.1%, respectively, p = 0.03). Length of time between the surgery and survey completion differed between the amputation and TPLO groups [925 (659–2248) days and 1494 (1037–2288) days, respectively, p < 0.01]. Owners’ assessments of pain severity prior to surgery on a 1 (none) to 4 (severe) scale were similar in the amputation [median 3 (2–4)] and TPLO [median 3 (3–4)] groups (p = 0.85). Appendix SA provides frequency distributions for answers to survey questions.

       Scale construction

      Overall, seven behavioral variables were identified that distinguished PAP from post-TPLO pain at p < 0.2 (Table 1). In multivariable logistic regression, four behaviors were independent predictors of PAP (p < 0.05; Fig. 2 & Table 2). Following adjustment of referent categories to make all score values positive, and with collapse of adjoining categories if coefficients were similar, these variables comprised the CAMPPAIN scale (Table 3). The scale included: 1) presence and severity of restlessness or difficulty sleeping, 2) presence and frequency of episodes of panic or anxiety, 3) presence and frequency of episodes of sudden vocalization, and 4) presence or absence of compulsive grooming of the residual limb. The score functioned on a 0 to 10 scale, with greater scores indicating greater likelihood of PAP. Score AUROC was 0.70 (95% confidence interval = 0.63–0.78) with good calibration (Hosmer–Lemeshow statistic p = 0.82). The risk probability of PAP could be calculated from the score as follows (where p = risk probability and R = logit p):
      PAP score: p = exp(R)/[1 + exp(R)]


      R = (1.021616 ∗ score) – 2.02185


      Table 1Postoperative behaviors differentiating postamputation pain (PAP, n = 66) from pain after tibial plateau-leveling osteotomy (TPLO, n = 139) in dogs determined by univariable analysis. A p value < 0.2 indicates the behavior was significantly more likely to occur in dogs that underwent amputation. CI, confidence interval
      BehaviorOdds ratio95% CIp
      Pain behavior triggered by startling or excitement3.701.16–11.780.025
      Hyporexia (moderate or severe)2.381.03–5.470.043
      Restlessness or difficulty sleeping (on a daily basis)3.471.45–8.310.005
      Episodes of anxiety or panic (on a monthly or more frequent basis)4.631.75–12.230.002
      Episodes of spontaneous, paroxysmal vocalization (on a monthly or more frequent basis)2.611.15–5.930.022
      Vocalization when residual limb touched2.220.69–7.160.17
      Compulsive grooming (licking or pulling of hair) at residual limb6.570.67–64.420.106
      Figure 2
      Figure 2Frequency distribution of behaviors [(a) restlessness, (b) panic or anxiety, (c) vocalization and (d) compulsive grooming] in survey responses from owners of dogs undergoing amputation (n = 66) and dogs undergoing tibial plateau-leveling osteotomy (TPLO, n = 139). These four behavioral variables comprise the Canine Postamputation Pain (CAMPPAIN) scale.
      Table 2Model coefficients from the multivariable model containing the four behavioral variables translated into the Canine Postamputation Pain (CAMPPAIN) scale, with the ‘never, or unable to recall’ category set as the referent category for all variables. CI, confidence interval
      Behavioral variableCoefficients95% CIp
      Restlessness or difficulty sleeping

       Occasional

       Weekly

       Daily


      0–0.59

      –0.65

      0.93


      –1.48 to 0.29

      –2.04 to 0.74

      –0.24 to 2.10
      0.05





      Episodes of anxiety or panic

       Rare

       Monthly

       Weekly

       Daily


      –0.12

      2.63

      0.79

      1.02


      –1.4 to 1.16

      0.28 to 4.99

      –0.96 to 2.54

      –1.32 to 3.35
      0.04







      Episodes of spontaneous, paroxysmal vocalization (on a monthly or more frequent basis)

       Rare

       Monthly

       Weekly

       Daily


      –0.04

      2.78

      –0.78

      –0.14


      –1.24 to 1.16

      0.45 to 5.12

      –2.53 to 0.98

      –1.83 to 1.55
      0.04







      Compulsive grooming (licking or pulling of hair) at residual limb2.970.66 to 5.27<0.01
      Table 3The Canine Postamputation Pain (CAMPPAIN) scale. Subtotals from each column are summed to obtain a total score, with a maximum possible total score of 10. A score ≥ 2 is 92% specific, 36% sensitive for the presence of postamputation pain
      BehaviorOccurrencePoints
      Restlessness/difficulty sleepingNever or can’t recall

      0.5 points
      Occasionally or weekly

      0 points
      Daily

      1.5 points


      ___ / 2 points
      Episodes of anxiety or panicNever, rarely, or can’t recall

      0 points
      Occasionally

      2.5 points
      Weekly or daily

      1 point


      ___ / 3.5 points
      VocalizationNever, rarely, or can’t recall

      0.5 points
      Monthly

      3 points
      Weekly or daily

      0 points


      ___ / 3.5 points
      Compulsive grooming of residual limbBehavior absent or can’t recall

      0 points
      Behavior noted

      3 points


      ___ / 3 points
      TOTAL

      ___ / 10 points
      A score of 2 corresponded to a risk probability of 0.5. Taking a score of ≥ 2 to indicate the presence of PAP gave a score specificity of 92.1% and sensitivity of 36.4%. When this criterion was used to diagnose PAP, prevalence was 36.4% (24/66) and 7.9% (11/139) in the amputation and TPLO groups, respectively.

      Discussion

      The authors were unable to find any previous publications describing specific behavioral changes that differentiate dogs that had undergone amputation from a comparison group that had undergone a common orthopedic surgery. Although previous studies have described behavioral changes in canine amputees, these behaviors may have been reported by owners of dogs undergoing other major surgeries (
      • Kirpensteijn J.
      • van den Bos R.
      • Endenburg N.
      Adaptation of dogs to the amputation of a limb and their owners' satisfaction with the procedure.
      ;
      • Menchetti M.
      • Gandini G.
      • Gallucci A.
      • et al.
      Approaching phantom complex after limb amputation in the canine species.
      ;
      • Ferreira M.G.
      • Antunes A.V.
      • Pascoli A.L.
      • et al.
      Neuropathic pain and prognosis of a dog submitted to limb amputation after diagnosis of soft tissue sarcoma - case report.
      ). The identification of behaviors that are more likely postamputation than post-TPLO suggests a pathophysiological process more common in amputees. Although the reason for such behaviors may never be certain, the authors suggest pain is most plausible. Based on experience with other veterinary patients and on descriptions by human patients, these behaviors appear to reflect neuropathic pain specifically.
      Assuming a score ≥ 2 to indicate PAP, prevalence was approximately 36% and 8% in the amputation and TPLO groups, respectively. Several possible explanations for the 8% prevalence in the TPLO group exist. One possibility is that the wording of some questions may have been too vague and led owners of dogs that had undergone TPLO to erroneously believe they had witnessed a behavior. For instance, in the authors’ experience, vocalization quality after amputation is always high-pitched and unprovoked, beginning and ending suddenly in an otherwise apparently comfortable dog (illustrated in the Video Clip S1), a characterization that was not clarified in the survey. By contrast, post-TPLO vocalization is continuous and/or triggered by palpation, weight bearing or movement. The second possibility is that some dogs experienced neuropathic pain post-TPLO. Although macroscopic nerves are not intentionally severed during TPLO as they are during amputation, cutting tissue during any surgery inevitably damages microscopic nerves. Indeed, suspected neuropathic pain has been documented in 39% of humans 1 month after total knee arthroplasty (
      • Fitzsimmons M.
      • Carr E.
      • Woodhouse L.
      • Bostick G.P.
      Development and persistence of suspected neuropathic pain after total knee arthroplasty in individuals with osteoarthritis.
      ). A third possibility is that nociceptive pain was present in both groups.
      Residual limb pain in humans can have a nociceptive component resulting from infection, failure of incisional healing, heterotopic ossification/osteophytes, edema, vascular insufficiency and myofascial pain/trigger points, among many other problems; such lesions may also have been present in dogs post-TPLO but went undiagnosed, or were diagnosed elsewhere (
      • Wolff A.
      • Vanduynhoven E.
      • van Kleef M.
      • et al.
      Phantom pain.
      ;
      • Clarke C.
      • Lindsay D.R.
      • Pyati S.
      • Buchheit T.
      Residual limb pain is not a diagnosis: a proposed algorithm to classify postamputation pain.
      ;
      • O'Reilly M.A.R.
      • O'Reilly P.M.R.
      • O'Reilly H.M.R.
      • et al.
      High-resolution ultrasound findings in the symptomatic residual limbs of amputees.
      ;
      • Buchheit T.
      • Van de Ven T.
      • Hsia H.L.J.
      • et al.
      Pain phenotypes and associated clinical risk factors following traumatic amputation: results from veterans integrated pain evaluation research (VIPER).
      ). In humans, some lesions can only be identified using diagnostic imaging (
      • O'Reilly M.A.R.
      • O'Reilly P.M.R.
      • O'Reilly H.M.R.
      • et al.
      High-resolution ultrasound findings in the symptomatic residual limbs of amputees.
      ). Some of the dogs developed infection post-TPLO, but infection and behavioral changes were not associated in all cases.
      The most common cause of neuropathic RLP is neuroma, a non-neoplastic proliferation of regenerating axons, Schwann cells and connective tissue that can develop as soon as 1 month after nerve transection and can be identified using diagnostic imaging (
      • O'Reilly M.A.R.
      • O'Reilly P.M.R.
      • O'Reilly H.M.R.
      • et al.
      High-resolution ultrasound findings in the symptomatic residual limbs of amputees.
      ;
      • Oliveira K.M.C.
      • Pindur L.
      • Han Z.
      • et al.
      Time course of traumatic neuroma development.
      ). Neuromas have been identified in tissue removed from the residual tail of dogs that presented for severe self-mutilation after tail docking (
      • Gross T.L.
      • Carr S.H.
      Amputation neuroma of docked tails in dogs.
      ). In humans, another cause of neuropathic RLP is sympathetically maintained pain, also called complex regional pain syndromes (CRPS) I and II (
      • Schlereth T.
      • Birklein F.
      The sympathetic nervous system and pain.
      ;
      • Knudsen L.F.
      • Terkelsen A.J.
      • Drummond P.D.
      • Birklein F.
      Complex regional pain syndrome: a focus on the autonomic nervous system.
      ). Although CRPS appears to occur only in humans, the sympathetic nervous system plays an important role in the pathophysiology of some pain states (
      • Schlereth T.
      • Birklein F.
      The sympathetic nervous system and pain.
      ). Following nerve injury, abnormal expression of functional adrenoceptors on peripheral nociceptors is stimulated; sympathetic activation (e.g., from stress) induces the release of norepinephrine, which then directly activates α-receptors on neurons (
      • Sato J.
      • Perl E.R.
      Adrenergic excitation of cutaneous pain receptors induced by peripheral nerve injury.
      ;
      • Schlereth T.
      • Birklein F.
      The sympathetic nervous system and pain.
      ). Peripheral nerve transection also stimulates sprouting of noradrenergic neurons into dorsal root ganglia, where they contact the cell bodies of the transected nerves (
      • McLachlan E.M.
      • Jänig W.
      • Devor M.
      • Michaelis M.
      Peripheral nerve injury triggers noradrenergic sprouting within dorsal root ganglia.
      ). In the univariable logistic regression model, pain triggered by startling or excitement (e.g., when a dog saw its owner or the doorbell rang) was significantly more likely to occur postamputation. Although this behavior was not retained in the multivariable logistic regression model, it deserves further investigation. No other features of CRPS were noted in the medical records or owner comments.
      Phantom limb pain may not be unique to humans, either. The most widely posited theory of the role of the CNS in PLP is the cortical remapping theory. This theory states that, following peripheral nerve transection or amputation, the somatosensory and motor cortices undergo massive reorganization such that a cortical region that formerly received projection neurons from a limb may begin to receive input from adjacent cortical regions (
      • Collins K.L.
      • Russell H.G.
      • Schumacher P.J.
      • et al.
      A review of current theories and treatments for phantom limb pain.
      ). Although cortical remapping is beyond the scope of this study, this reorganization has been documented in animals, including nonhuman primates, rodents and cats, after nerve injury (
      • Kaas J.H.
      • Merzenich M.M.
      • Killackey H.P.
      The reorganization of somatosensory cortex following peripheral nerve damage in adult and developing mammals.
      ;
      • Merzenich M.M.
      • Kaas J.H.
      • Wall J.
      • et al.
      Topographic reorganization of somatosensory cortical areas 3b and 1 in adult monkeys following restricted deafferentation.
      ,
      • Merzenich M.M.
      • Kaas J.H.
      • Wall J.T.
      • et al.
      Progression of change following median nerve section in the cortical representation of the hand in areas 3b and 1 in adult owl and squirrel monkeys.
      ).
      The behaviors comprising the CAMPPAIN scale have been identified in previous research on PAP (
      • Kirpensteijn J.
      • van den Bos R.
      • Endenburg N.
      Adaptation of dogs to the amputation of a limb and their owners' satisfaction with the procedure.
      ;
      • Menchetti M.
      • Gandini G.
      • Gallucci A.
      • et al.
      Approaching phantom complex after limb amputation in the canine species.
      ;
      • Ferreira M.G.
      • Antunes A.V.
      • Pascoli A.L.
      • et al.
      Neuropathic pain and prognosis of a dog submitted to limb amputation after diagnosis of soft tissue sarcoma - case report.
      ). Humans describe neuropathic RLP and PLP as sharp, tingling, shooting, stabbing, throbbing, aching, burning, shocking or some combination of these (
      • Ehde D.M.
      • Czerniecki J.M.
      • Smith D.G.
      • et al.
      Chronic phantom sensations, phantom pain, residual limb pain, and other regional pain after lower limb amputation.
      ). In humans, the pain is usually intermittent and paroxysmal (rather than constant), occurring daily or weekly, with attacks lasting from seconds to hours in most patients (
      • Ehde D.M.
      • Czerniecki J.M.
      • Smith D.G.
      • et al.
      Chronic phantom sensations, phantom pain, residual limb pain, and other regional pain after lower limb amputation.
      ;
      • Desmond D.M.
      • Maclachlan M.
      Prevalence and characteristics of phantom limb pain and residual limb pain in the long term after upper limb amputation.
      ). In a previous PAP study, 53% of owners described their dogs’ pain as ‘sudden and transient’; none described it as ‘persistent’ (
      • Menchetti M.
      • Gandini G.
      • Gallucci A.
      • et al.
      Approaching phantom complex after limb amputation in the canine species.
      ). Sudden, severe pain could explain the spontaneous, paroxysmal vocalizations and episodes of panic or anxiety in canine amputees. These episodes appear to be caused by abnormal electrogenesis in the peripheral and CNS, caused by upregulation of voltage-gated sodium channels and numerous other molecular changes, which create sudden depolarizations called ectopic discharges (
      • Collins K.L.
      • Russell H.G.
      • Schumacher P.J.
      • et al.
      A review of current theories and treatments for phantom limb pain.
      ). In a small percentage of humans with PAP, the pain is constant (
      • Ehde D.M.
      • Czerniecki J.M.
      • Smith D.G.
      • et al.
      Chronic phantom sensations, phantom pain, residual limb pain, and other regional pain after lower limb amputation.
      ). Continuous pain in some dogs may be responsible for the restlessness and compulsive grooming observed by some owners. Notably, resentment of palpation of the residual limb was not retained in the multivariable model. Human amputees can suffer severe hyperesthesia and allodynia at the residual limb; however, palpation may identify areas of hypoesthesia (
      • Nikolajsen L.
      • Ilkjaer S.
      • Christensen J.H.
      • et al.
      Pain after amputation.
      ). Because pain may occur only during ectopic discharges, and areas of hypoesthesia may be present, clinicians should not rule out neuropathic pain based on lack of response to palpation. Physical examination is often unremarkable in human amputees (
      • Wolff A.
      • Vanduynhoven E.
      • van Kleef M.
      • et al.
      Phantom pain.
      ).
      The CAMPPAIN scale could serve several useful functions. It would establish diagnostic criteria for PAP in dogs, facilitating recognition of this syndrome by clinicians and prompting adjustment of pain management protocols. It may alert owners caring for amputees to the need for additional medical intervention. Finally, it may facilitate selection of homogenous clinical research populations for clinical trials targeted at determining risk factors for PAP in dogs and reducing incidence.
      Applying population-generated probabilities to an individual animal that can only experience a binary outcome always raises challenges. The CAMPPAIN scale cut-point of ≥ 2 for diagnosis optimizes specificity. On the one hand, this allows a high degree of confidence that PAP is present if the score indicates as such, while minimizing false positives. This is important if a PAP diagnosis would prompt a change in prescriptions, particularly of controlled substances. On the other hand, low sensitivity may lead to undertreatment of PAP; therefore, if a clinician has a high suspicion that a dog has PAP, it may be reasonable to treat for pain and monitor the dog’s response, regardless of the CAMPPAIN score.
      The definition of the relationship between a variable and an outcome event in the context of a multivariable model is the association between the two that remains after the influences of all other variables acting concurrently have been controlled. As such, the relationship between a variable and outcome event may differ substantially from that seen in a univariable setting, where only that variable is modelled. Because the PAP score is a direct reflection of the multivariable model that was used to develop it, which in turn was data-dependent, some of the scores for the different subcategories do not ascend according to intuitive expectations of their progression in a univariable context. A prospective study probably will lead to revision of the scale, with increasing frequency of a behavior increasing the score in that subcategory.
      The most important limitation of the present study is the retrospective nature and possible bias. More time passed between surgery and survey completion in the TPLO group; therefore, these responses may be less accurate. There was substantial passage of time between surgery and survey completion in both groups. Although this provided ample opportunity for owners to witness chronic behaviors, extended time may have contributed to recall bias. The ‘unknown/can’t remember’ response is often a problem with retrospective surveys, and the most conservative method of handling this without major data loss was making this response equivalent to a null value. Replacing this response with ‘missing’ would have resulted in loss of the entire survey response from the model and may have introduced bias. Owners of amputees may have been more likely to respond if their dog experienced problems after surgery than if their dog recovered uneventfully. Questionnaires may underestimate the frequency of behaviors because owners may not observe all episodes of such behaviors. Furthermore, model and scale performance characteristics are generally inflated when calculated on the population from which a scale is derived (
      • Hosmer D.W.
      • Lemeshow S.
      Applied Logistic Regression.
      ). Dogs in each group could not be matched based on signalment; the amputation and TPLO groups differed by both age and body weight. The authors could not find any publications demonstrating association between either of these characteristics and differences in the way neuropathic pain is displayed in dogs. However, increasing age has been associated with neuropathic pain in humans, and the possibility that the age difference, albeit relatively minor, may have introduced bias cannot be ruled out (
      • Fitzgerald M.
      • McKelvey R.
      Nerve injury and neuropathic pain - a question of age.
      ). The present study was single center, and the anesthetic/analgesic protocols for the two groups were not standardized. The incidence of PAP may be influenced by pain management practices, which vary from center to center, and between clinicians within the same center; consequently, these results may not be reflective of other centers. Certain drugs or locoregional techniques may decrease the risk of PAP. For all these reasons, a prospective study to externally validate the CAMPPAIN scale is required.

      Conclusions

      A 10 point scale was developed that could be simply calculated from readily observed behaviors in dogs that showed high specificity for pain after amputation. Following external validation, this scale may improve recognition of this syndrome and facilitate further clinical trials targeted at reducing incidence.

      Acknowledgements

      The authors thank Dr. Abigail Hooker, and Carol Frederick and Lucinda Bennett, licensed veterinary technicians, at the Cornell University College of Veterinary Medicine for assistance with data collection. This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

      Authors’ contributions

      JMB and DJL: study conception and design, data collection, manuscript preparation. KER and AKN: data collection, manuscript preparation. LC and RDG: study conception and design, manuscript preparation. GMH: study conception and design, statistical analysis, manuscript preparation. All authors read and approved the final version of the manuscript.

      Conflict of interest statement

      The authors declare no conflict of interest.

      Supporting Information

      The following are the Supplementary data to this article:

      References

        • Alviar M.J.M.
        • Hale T.
        • Dungca M.
        Pharmacologic interventions for treating phantom limb pain.
        Cochrane Database Syst Rev. 2016; 10CD006380
        • Buchheit T.
        • Van de Ven T.
        • Hsia H.L.J.
        • et al.
        Pain phenotypes and associated clinical risk factors following traumatic amputation: results from veterans integrated pain evaluation research (VIPER).
        Pain Med. 2016; 17: 149-161
        • Carberry C.A.
        • Harvey H.J.
        Owner satisfaction with limb amputation in dogs and cats.
        J Am Anim Hosp Assoc. 1987; 23: 227-232
        • Clarke C.
        • Lindsay D.R.
        • Pyati S.
        • Buchheit T.
        Residual limb pain is not a diagnosis: a proposed algorithm to classify postamputation pain.
        Clin J Pain. 2013; 29: 551-562
        • Collins K.L.
        • Russell H.G.
        • Schumacher P.J.
        • et al.
        A review of current theories and treatments for phantom limb pain.
        J Clin Invest. 2018; 128: 2168-2176
        • Desmond D.M.
        • Maclachlan M.
        Prevalence and characteristics of phantom limb pain and residual limb pain in the long term after upper limb amputation.
        Int J Rehabil Res. 2010; 33: 279-282
        • Dickerson V.M.
        • Coleman K.D.
        • Ogawa M.
        • et al.
        Outcomes of dogs undergoing limb amputation, owner satisfaction with limb amputation procedures, and owner perceptions regarding postsurgical adaptation: 64 cases (2005-2012).
        J Am Vet Med Assoc. 2015; 247: 786-792
        • Ehde D.M.
        • Czerniecki J.M.
        • Smith D.G.
        • et al.
        Chronic phantom sensations, phantom pain, residual limb pain, and other regional pain after lower limb amputation.
        Arch Phys Med Rehabil. 2000; 81: 1039-1044
        • Ferreira M.G.
        • Antunes A.V.
        • Pascoli A.L.
        • et al.
        Neuropathic pain and prognosis of a dog submitted to limb amputation after diagnosis of soft tissue sarcoma - case report.
        Arquivo Brasileiro De Medicina Veterinaria E Zootecnia. 2019; 71: 1116-1120
        • Fitzgerald M.
        • McKelvey R.
        Nerve injury and neuropathic pain - a question of age.
        Exp Neurol. 2016; 275: 296-302
        • Fitzsimmons M.
        • Carr E.
        • Woodhouse L.
        • Bostick G.P.
        Development and persistence of suspected neuropathic pain after total knee arthroplasty in individuals with osteoarthritis.
        PM R. 2018; 10: 903-909
        • Galindo-Zamora V.
        • von Babo V.
        • Eberle N.
        • et al.
        Kinetic, kinematic, magnetic resonance and owner evaluation of dogs before and after the amputation of a hind limb.
        BMC Vet Res. 2016; 12: 20
        • Gross T.L.
        • Carr S.H.
        Amputation neuroma of docked tails in dogs.
        Vet Pathol. 1990; 27: 61-62
        • Hanley M.A.
        • Jensen M.P.
        • Smith D.G.
        • et al.
        Preamputation pain and acute pain predict chronic pain after lower extremity amputation.
        J Pain. 2007; 8: 102-109
        • Hayes G.
        • Mathews K.
        • Doig G.
        • et al.
        The acute patient physiologic and laboratory evaluation (APPLE) score: a severity of illness stratification system for hospitalized dogs.
        J Vet Intern Med. 2010; 24: 1034-1047
        • Hosmer D.W.
        • Lemeshow S.
        Applied Logistic Regression.
        2nd edn. John Wiley & Sons Inc., USA2000: 91-116
        • Kaas J.H.
        • Merzenich M.M.
        • Killackey H.P.
        The reorganization of somatosensory cortex following peripheral nerve damage in adult and developing mammals.
        Annu Rev Neurosci. 1983; 6: 325-356
        • Kirpensteijn J.
        • van den Bos R.
        • Endenburg N.
        Adaptation of dogs to the amputation of a limb and their owners' satisfaction with the procedure.
        Vet Rec. 1999; 144: 115-118
        • Knudsen L.F.
        • Terkelsen A.J.
        • Drummond P.D.
        • Birklein F.
        Complex regional pain syndrome: a focus on the autonomic nervous system.
        Clin Auton Res. 2019; 29: 457-467
        • Kooijman C.M.
        • Dijkstra P.U.
        • Geertzen J.H.B.
        • et al.
        Phantom pain and phantom sensations in upper limb amputees: an epidemiological study.
        Pain. 2000; 87: 33-41
        • Kowaleski M.
        • Boudrieau R.
        • Pozzi A.
        Stifle joint.
        in: Tobias K.M. Johnston S.A. Veterinary Surgery: Small Animal. 2nd edn. Elsevier, USA2012: 906-998
        • Kuffler D.P.
        Origins of phantom limb pain.
        Mol Neurobiol. 2018; 55: 60-69
        • Lopez D.J.
        • VanDeventer G.M.
        • Krotscheck U.
        • et al.
        Retrospective study of factors associated with surgical site infection in dogs following tibial plateau leveling osteotomy.
        J Am Vet Med Assoc. 2018; 253: 315-321
        • Machin P.
        • Williams A.C.
        Stiff upper lip: coping strategies of World War II veterans with phantom limb pain.
        Clin J Pain. 1998; 14: 290-294
        • McLachlan E.M.
        • Jänig W.
        • Devor M.
        • Michaelis M.
        Peripheral nerve injury triggers noradrenergic sprouting within dorsal root ganglia.
        Nature. 1993; 363: 543-546
        • Menchetti M.
        • Gandini G.
        • Gallucci A.
        • et al.
        Approaching phantom complex after limb amputation in the canine species.
        J Vet Behav. 2017; 22: 24-28
        • Merzenich M.M.
        • Kaas J.H.
        • Wall J.
        • et al.
        Topographic reorganization of somatosensory cortical areas 3b and 1 in adult monkeys following restricted deafferentation.
        Neuroscience. 1983; 8: 33-55
        • Merzenich M.M.
        • Kaas J.H.
        • Wall J.T.
        • et al.
        Progression of change following median nerve section in the cortical representation of the hand in areas 3b and 1 in adult owl and squirrel monkeys.
        Neuroscience. 1983; 10: 639-665
        • Nikolajsen L.
        • Ilkjaer S.
        • Christensen J.H.
        • et al.
        Pain after amputation.
        Br J Anaesth. 1998; 81: 486
        • O'Reilly M.A.R.
        • O'Reilly P.M.R.
        • O'Reilly H.M.R.
        • et al.
        High-resolution ultrasound findings in the symptomatic residual limbs of amputees.
        Mil Med. 2013; 178: 1291-1297
        • Oliveira K.M.C.
        • Pindur L.
        • Han Z.
        • et al.
        Time course of traumatic neuroma development.
        PLoS One. 2018; 13e0200548
        • Padovani M.T.
        • Martins M.R.I.
        • Venâncio A.
        • Forni J.E.N.
        Anxiety, depression and quality of life in individuals with phantom limb pain.
        Acta Ortop Bras. 2015; 23: 107-110
        • Paré A.
        La maniere de traicter les playes faictes tant par hacquebutes: que par fleches: & les accidentz d'icelles, comme fractures & caries des os, grangrene & mortification; avec les pourtraictz des instrumentz necessaires pour leur curation. Et La methode de curer les combustions principalement faictes par la pouldre a canon.
        Arnoul L'Angelie, 1552 (Paris U6 - ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&rfr_id=info:sid/summon.serialssolutions.com&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=book&rft.title=La+maniere+de+traicter+les+playes+faictes+tant+par+hacquebutes&rft.au=Paré,+Ambroise&rft.series=French+books+before+1601&rft.date=1552-01-01&rft.pub=Arnoul+L'Angelie&rft.volume=r.334+item+4&rft.externalDocID=b13560293&paramdict=en-US U7 - Book)
        • Sato J.
        • Perl E.R.
        Adrenergic excitation of cutaneous pain receptors induced by peripheral nerve injury.
        Science. 1991; 251: 1608-1610
        • Schlereth T.
        • Birklein F.
        The sympathetic nervous system and pain.
        Neuromolec Med. 2008; 10: 141-147
        • Seguin B.
        • Weigel J.
        Amputations.
        in: Tobias K.M. Johnston S.A. Veterinary Surgery: Small Animal. 2nd edn. Elsevier, USA2012: 1029-1036
        • Task Force on Taxonomy
        IASP Terminology. International Association for the Study of Pain (IASP).
        2017
        • von Elm E.
        • Altman D.G.
        • Egger M.
        • et al.
        The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies.
        Lancet. 2007; 370: 1453-1457
        • Wolff A.
        • Vanduynhoven E.
        • van Kleef M.
        • et al.
        Phantom pain.
        Pain Pract. 2011; 11: 403-413
        • Ziegler-Graham K.
        • MacKenzie E.J.
        • Ephraim P.L.
        • et al.
        Estimating the prevalence of limb loss in the United States: 2005 to 2050.
        Arch Phys Med Rehabil. 2008; 89: 422-429