Over the past decade, pain management has become an integral part of veterinary medicine as new drugs and treatment modalities have become available. Valid measures of chronic pain are needed not only to help effectively treat pain in individual patients, but also to enable the testing of reliability and effectiveness of various pain interventions.
Although signs of pain can be assessed by means of veterinary evaluation, owner observation, or gait analysis, each method has limitations and strengths. Gait analysis measurements, although more objective, are time-consuming, require specialized equipment, and assess the patient only at a specific time point in a hospital or laboratory environment. Veterinary evaluations of chronic pain are subjective and can be done only at a specific time point, with the potential for subtle clinical signs being masked during in-clinic examinations because of a dog's excitement or stress responses. Owner evaluation of chronic pain, although also subjective, has the advantage of integrating observations over an extended period of time as dogs go about their regular activities in the home setting.
Several research programs have focused on developing instruments for owner assessment of chronic pain in dogs.1–4 Wiseman-Orr et al1,2 assessed the impact of chronic pain on health-related quality of life. Yazbek and Fantoni4 described the development of a questionnaire for evaluation of pain in dogs with cancer; differences in scores were detected between healthy dogs and dogs with cancer, but no treatment was evaluated. The Helsinki chronic pain index was developed and tested in a study3 of 34 dogs, in which significant differences in mean pain scores were found between placebo and carprofen treatment groups.
The CBPI is a publically available owner-completed questionnaire5 designed to quantify the severity of chronic pain and its impact on routine activities in companion dogs. The questionnaire, which was developed for owners to complete, includes questions pertaining to the severity of pain and the degree to which pain interferes with the dog's typical activities; scores assigned by the owner are used to generate an overall PSS and PIS.6,7 The CBPI was created by use of standard methods for the stepwise development of a health-assessment questionnaire.8–10 Development and validation of the CBPI instrument have been described elsewhere.6,7,11,12
In a blinded, randomized clinical trial12 to evaluate use of the CBPI to detect differences between carprofen- and placebo-treated dogs with osteoarthritis, owners completed the questionnaire for evaluation of their dogs before the start of treatment and after 14 days of treatment. The questionnaire was able to distinguish between the effect of the placebo and the effect of carprofen treatment, and results demonstrated that carprofen was more effective in reducing the pain score than placebo treatment. The median PSS and PIS were significantly different between the placebo and carprofen groups, indicating that the CBPI was able to detect improvements in pain scores in both groups as well as a difference in improvement between the groups. Although the treatment effect of carprofen was successfully detected, a limitation of that analysis was that it provided little information about treatment success or failure in each study dog. Large improvements in a small number of dogs within a group could potentially mask a lack of improvement in the remaining dogs in the group.
In the study reported here, we sought to explore an alternative approach to data analysis for studies of drug effectiveness in which the CBPI is used. In this approach, rather than comparing the overall mean or median differences in scores between groups, the goal is to assess whether the treatment has a measurable effect for individual animals. The criteria for successful treatment of an individual patient are predefined, so that success or failure of the treatment in each animal can be determined at study completion. The number of treatment successes and failures in each group (generally animals that receive an active agent vs those administered a placebo) can then be compared. This method has the advantage of reflecting how likely a treatment is to be effective in an individual patient, rather than in a group of patients. In pivotal clinical studies evaluating NSAIDs for the treatment of pain in dogs, the FDA Center for Veterinary Medicine13 has required this approach.
Several drugs have been approved for the control of pain and inflammation associated with osteoarthritis in dogs during the past 10 years,13 and pharmaceutical companies are continuing to develop novel treatments.14–16 The purpose of the study reported here was to evaluate the power of the CBPI for use in detecting differences among treatment groups in studies of osteoarthritis in dogs, when the data are analyzed on the basis of individual treatment success, rather than comparing mean or median data, by analyzing CBPI data for dogs treated with carprofen or a placebo, assessing the effects of various inclusion requirements and predetermined criteria for treatment success or failure at the individual patient level, and then evaluating the ability of the instrument to detect differences between treatment groups.
Materials and Methods
Animals—Data from dogs in a previous study12 conducted by our group were analyzed retrospectively along with data collected in the same manner from an additional set of dogs for the present study. Dogs with a body weight > 8 kg and medical history, clinical signs, physical examination findings, and radiographic findings consistent with osteoarthritis were eligible for inclusion in each study. Only dogs with newly diagnosed osteoarthritis or dogs that had received no previous treatment for osteoarthritis were enrolled. Dogs that had any concomitant disease or were receiving medications were excluded.2
All owners received a detailed written description of the protocol and signed a consent form before their dogs were evaluated for possible study inclusion. All dogs that met the described eligibility criteria and whose owners agreed to participate were enrolled. No minimum PSS or PIS was specified. Owners had the option to withdraw a dog from the study and seek alternative treatment at any time. The study protocols were approved by an institutional animal care and use committee.
CBPI—Briefly, the CBPI consists of questions to be completed by a dog's owner and is separated into 2 sections.5,7 The first section is used to calculate the PSS and consists of 4 questions that ask owners to rate their impression of the dog's pain at its worst, at its least, on average, and “right now” (at the time the form is completed), on a scale of 0 (no pain) to 10 (extreme pain). The PSS is calculated as the mean of these 4 scores. The second section is used to calculate the PIS and includes 6 questions that ask owners to rate the degree to which they believe pain interferes with their dog's general activity, enjoyment of life, and actions such as the ability to rise to standing from recumbency, walking, running, and climbing (eg, stairs or curbs) on a scale of 0 (does not interfere) to 10 (completely interferes). The PIS is calculated as the mean of these 6 scores. Overall quality of life is also rated on a 5-point scale from poor to excellent. This is a stand-alone question that is not used in the calculation of pain scores.
Study design—The previous study12 and the present study were blinded, randomized, controlled clinical trials with identical designs, and no dogs were enrolled in both studies. The practice of pooling data from 2 or more independent data sets generated through identical study designs was used.17–19
Calculations for the previous study12 indicated that a sample size of 29 dogs/group was required to achieve 80% power to detect a 30% difference between treatment groups (SD, 40%; 2-sided P = 0.05) for changes in CBPI scores. To compensate for potential dropouts, it was determined that 70 dogs would be randomly assigned to receive either carprofen or a placebo (35 dogs/group). A larger sample size (80 dogs [40/group]) was used to collect new data in the same manner for the present study. Therefore, data from 150 dogs were included in the analyses.
Randomization procedures and blinding were the same for the present study as for the previous study12 and have been described in detail elsewhere. Briefly, pharmacy personnel matched study animal identification numbers with a randomization sequence, packaged either the carprofen or placebo formulation (a combination of lactose and microcrystalline cellulose) into identical blister packs, and dispensed the packs with instructions to owners. All treatments were administered in a similar manner (PO, q 24 h); carprofen was administered at the labeled dose (4.4 mg/kg, PO, q 24 h). All study personnel and dog owners were blinded to treatment assignments.
Dogs were screened for eligibility 7 to 10 days prior to random group assignment, and the day 0 (baseline) CBPI was completed by owners at that time to familiarize them with use of the instrument and to minimize regression to the mean associated with owners seeking out a study when their dog's condition was at its worst.
The same owner who completed the CBPI at baseline completed a second CBPI on day 14. Baseline CBPI scores were not available for owners to review as they were completing the day 14 CBPI. Owners returned their dogs to the clinic for a follow-up visit on day 14. After completion of the study, all owners were provided a 2-week supply of carprofen for administration at the previously described dosage, and information was given on options for continued treatment.
Statistical analysis—For analysis purposes, it was assumed that a relevant treatment success in each dog in the combined data set should include decreases in both the PSS and the PIS. Therefore, treatment success was defined as a reduction of at least 1 in both the PSS and PIS on day 14, compared with the respective baseline values. Use of this definition of treatment success required that the criterion for inclusion in the analysis was set at a mean score ≥ 1 for each variable at baseline. Thus, dogs that were very mildly affected were removed from the analysis, making the definition of successful treatment more robust and conforming to the type of study design used by the FDA Center for Veterinary Medicine, where minimum levels of pain (determined by various means) are required before a dog can be enrolled in a pivotal clinical trial.
The analysis performed on this data set explored the power of defining treatment success as a reduction of 1, 2, or 3 in either or both the PSS and PIS as well as how setting the inclusion criteria at baseline to a PSS and PIS ≥ 1, 2, or 3 affected the power of the statistical analysis to detect differences between the placebo and carprofen treatment.
Due to the nature of osteoarthritis and the method of pain scoring, baseline PSS and PIS were expected to be highly correlated. To test this assumption, the Pearson correlation coefficient and a corresponding P value were generated for these variables.
The first step in the analysis of the combined data set was to define minimum baseline PSS and PIS for inclusion in the analysis. To evaluate the effect of setting various minimum baseline scores as inclusion criteria, 3 populations were constructed. These comprised dogs with a PSS ≥ 1 and PIS ≥ 1 at baseline (population 1), PSS ≥ 2 and PIS ≥ 2 at baseline (population 2), or PSS ≥ 3 and PIS ≥ 3 at baseline (population 3).
The second step in the analysis was to define success criteria as various reductions in PSS and PIS on day 14, compared with the respective baseline scores. For population 1, one possible success criterion was evaluated: a reduction ≥ 1 in both the PSS and PIS. For population 2, this same criterion was investigated along with 3 others: reduction ≥ 1 in PSS and ≥ 2 in PIS, reduction ≥ 2 in PSS and ≥ 1 in PIS, or reduction ≥ 2 in both PSS and PIS. For population 3, the same 4 possible success criteria described for populations 1 and 2 were evaluated in addition to 5 others: reduction ≥ 1 in PSS and ≥ 3 in PIS, reduction ≥ 2 in PSS and ≥ 3 in PIS, reduction ≥ 3 in PSS and ≥ 1 in PIS, reduction ≥ 3 in PSS and ≥ 2 in PIS, or reduction ≥ 3 in both PSS and PIS.
For each population and success criteria, the number and percentage of treatment successes and failures were summarized by treatment group. Possible differences between treatment groups were evaluated with the χ2 test. For each definition of success within each population, power was calculated by means of a continuity-corrected 2-sided z test, with α = 0.05.
To evaluate these same data on the basis of median scores as done previously,12 changes in median PSS and PIS from baseline to day 14 for the carprofen- and placebo-treated dogs in populations 1, 2, and 3 were calculated by ANOVA of the ranked scores. All statistical analyses were conducted with commercially available software.a
Results
All dogs that were enrolled in the clinical trials completed them. One hundred forty-nine were included in the final analysis. When all dogs were evaluated, regardless of baseline score, the baseline PSS and PIS had a strong positive correlation (r = 0.752; P < 0.001).
Population 1—Population 1 included all dogs with a PSS ≥ 1 and PIS ≥ 1 at baseline. With success defined as a reduction ≥ 1 in both PSS and PIS, treatment was classified as successful in 38 of 70 (54.3%) carprofen-treated dogs and 22 of 67 (32.8%) placebo-treated dogs (Table 1). The success rate was significantly (P = 0.011) higher for dogs in the carprofen group.
Treatment response assessments (No. [%]) as determined by use of the CBPI in a group of dogs with osteoarthritis (population 1; n = 137) for which inclusion criteria were set as PSS ≥ 1 and PIS ≥ 1 at baseline (day 0) and the treatment success criterion was predefined as a reduction ≥ 1 in both the PSS and PIS on day 14.
| Outcome | Carprofen | Placebo | P value | Power |
|---|---|---|---|---|
| Success | 38 (54.3) | 22 (32.8) | 0.011 | 45.2 |
| Failure | 32 (45.7) | 45 (67.2) | — | — |
Dogs received carprofen or a placebo administered by owners beginning on day 1 in randomized, blinded, controlled clinical trials. All treatments were administered in a similar manner (PO, q 24 h); carprofen was administered at the labeled dose (4.4 mg/kg, PO, q 24 h). P values were derived with the χ2 test.
— = Not evaluated separately.
Population 2—For population 2 (116 dogs with PSS ≥ 2 and PIS ≥ 2 at baseline), when success was defined as a reduction ≥ 1 in both the PSS and PIS, treatment was classified as successful in 33 of 57 (57.9%) carprofen-treated dogs and 22 of 59 (37.3%) placebo-treated dogs (Table 2). The success rate was significantly (P = 0.026) higher for the carprofen group.
Treatment response assessments (No. [%]) in a group of dogs with osteoarthritis (population 2; n = 116) for which inclusion criteria were set as PSS ≥ 2 and PIS ≥ 2 at baseline and various treatment success criteria were predefined on the basis of changes in the PSS and PIS on day 14.
| Success criteria (reduction in score) | Outcome | Carprofen | Placebo | P value | Power |
|---|---|---|---|---|---|
| PSS ≥ 1; PIS ≥ 1 | Success | 33 (57.9) | 22 (37.3) | 0.026 | 54.0 |
| Failure | 24 (42.1) | 37 (62.7) | — | — | |
| PSS ≥ 1; PIS ≥ 2 | Success | 26 (45.6) | 14 (23.7) | 0.013 | 62.8 |
| Failure | 31 (54.4) | 45 (76.3) | — | — | |
| PSS ≥ 2; PIS ≥ 1 | Success | 20 (35.1) | 13 (22.0) | 0.119 | 26.9 |
| Failure | 37 (64.9) | 46 (78.0) | — | — | |
| PSS ≥ 2; PIS ≥ 2 | Success | 18 (31.6) | 9 (15.3) | 0.038 | 45.9 |
| Failure | 39 (68.4) | 50 (84.7) | — | — |
See Table 1 for key.
When success was defined as a reduction ≥ 1 in PSS and ≥ 2 in PIS, treatment was classified as successful in 26 of 57 (45.6%) carprofen-treated dogs and 14 of 59 (23.7%) placebo-treated dogs. The success rate was found to be significantly (P = 0.013) higher for the carprofen group.
When success was defined as reduction ≥ 2 in PSS and ≥ 1 in PIS, treatment was classified as successful for 20 of 57 (35.1%) carprofen-treated dogs and 13 of 59 (22.0%) placebo-treated dogs. This difference was nonsignificant (P = 0.119).
When success was defined as a reduction ≥ 2 in both the PSS and PIS, treatment was classified as successful in 18 of 57 (31.6%) carprofen-treated dogs and 9 of 59 (15.3%) placebo-treated dogs. Although the success rate was low for both groups, it was significantly (P = 0.038) higher for the carprofen group.
Population 3—Population 3 included 87 dogs with a PSS ≥ 3 and PIS ≥ 3 at baseline. Only 2 of the 9 definitions of treatment success yielded success rates ≥ 50% in the carprofen-treated dogs (Table 3). When treatment success was defined as a reduction ≥ 1 in both the PSS and PIS, treatment was classified as successful in 27 of 43 (62.8%) carprofen-treated dogs, compared with 17 of 44 (38.6%) placebo-treated dogs (P = 0.024). Treatment was considered successful in 22 of 43 (51.2%) carprofen-treated dogs and 11 of 44 (25.0%) placebo-treated dogs when success was defined as a reduction ≥ 1 in PSS and ≥ 2 in PIS (P = 0.012).
Treatment response assessments (No. [%]) in a group of dogs with osteoarthritis (population 3; n = 87) for which inclusion criteria were set as PSS ≥ 3 and PIS ≥ 3 at baseline and various treatment success criteria were predefined on the basis of changes in the PSS and PIS on day 14.
| Success criteria (reduction in score) | Outcome | Carprofen | Placebo | P value | Power |
|---|---|---|---|---|---|
| PSS ≥ 1; PIS ≥ 1 | Success | 27 (62.8) | 17 (38.6) | 0.024 | 54.9 |
| Failure | 16 (37.2) | 27 (61.4) | — | — | |
| PSS ≥ 1; PIS ≥ 2 | Success | 22 (51.2) | 11 (25.0) | 0.012 | 64.7 |
| Failure | 21 (48.8) | 33 (75.0) | — | — | |
| PSS ≥ 1; PIS ≥ 3 | Success | 10 (23.3) | 5 (11.4) | 0.142 | 22.3 |
| Failure | 33 (76.7) | 39 (88.6) | — | — | |
| PSS ≥ 2; PIS ≥ 1 | Success | 17 (39.5) | 9 (20.5) | 0.052 | 40.6 |
| Failure | 26 (60.5) | 35 (79.5) | — | — | |
| PSS ≥ 2; PIS ≥ 2 | Success | 16 (37.2) | 7 (15.9) | 0.024 | 53.9 |
| Failure | 27 (62.8) | 37 (84.1) | — | — | |
| PSS ≥ 2; PIS ≥ 3 | Success | 8 (18.6) | 3 (6.8) | 0.098 | 26.7 |
| Failure | 35 (81.4) | 41 (93.2) | — | — | |
| PSS ≥ 3; PIS ≥ 1 | Success | 11 (25.6) | 2 (4.5) | 0.006 | 74.4 |
| Failure | 32 (74.4) | 42 (95.5) | — | — | |
| PSS ≥ 3; PIS ≥ 2 | Success | 10 (23.3) | 2 (4.5) | 0.011 | 65.0 |
| Failure | 33 (76.7) | 42 (95.5) | — | — | |
| PSS ≥ 3; PIS ≥ 3 | Success | 7 (16.3) | 1 (2.3) | 0.024 | 46.4 |
| Failure | 36 (83.7) | 43 (97.7) | — | — |
See Table 1 for key.
Power analysis and comparison with median group scores—Results of the power analyses were summarized for each of the study populations (Tables 1–3). For population 2, the success criteria that yielded the highest power (62.8) were reductions ≥ 1 in PSS and ≥ 2 in PIS. For population 3, a similar power (64.7) was obtained with these same success criteria, but some of the alternate success criteria yielded similar or higher power values.
As was previously found, when data were evaluated on the basis of median group PSS and PIS, the change in scores from baseline to day 14 was significantly different between carprofen- and placebo-treated dogs from populations 1, 2, and 3 (Table 4). In each comparison, the change in median score indicated greater improvement in carprofen-treated dogs than in placebo-treated dogs.
Median (range) change in PSS and PIS in 3 groups of dogs with osteoarthritis (populations 1, 2, and 3) on day 14 of treatment with carprofen or a placebo, compared with baseline scores (negative values indicate improvement).
| Change in score | |||||||
|---|---|---|---|---|---|---|---|
| PSS | PIS | ||||||
| Population | No. of dogs | Carprofen | Placebo | P value | Carprofen | Placebo | P value |
| 1 | 137 | −1.25 (−6 to 2.75) | −0.50 (−7.25 to 1.75) | 0.002 | −1.67 (−7 to 1.34) | −0.50 (−8.34 to 3) | < 0.001 |
| 2 | 116 | −1.50 (−6 to 2.75) | −0.75 (−7.25 to 1.75) | 0.009 | −1.84 (−7 to 1.34) | −0.50 (−8.34 to 3) | < 0.001 |
| 3 | 87 | −1.25 (−6 to 2.75) | −0.50 (−7.25 to 1.75) | 0.001 | −1.58 (−7 to 1.34) | −0.33 (−8.34 to 3) | < 0.001 |
Populations of dogs were defined on the basis of various inclusion criteria at baseline (PSS and PIS ≥ 1 for population 1, PSS and PIS ≥ 2 for population 2, and PSS and PIS ≥ 3 for population 3). P values were generated by analysis of variance of ranked values.
Discussion
Consistent with prior reports,11,12 the CBPI in the present study, involving the use of predefined definitions of treatment success in individual patients, detected differences in pain score changes that would be expected in a trial performed to compare an active NSAID treatment with a placebo treatment in dogs. The results of pooled data analysis (including independent data sets collected in 2 clinical trials with identical designs) indicated that the CBPI could also be used to detect differences between treatments when a PSS and PIS ≥ 1 were required for study inclusion, and treatment success was defined as a reduction of at least 1 in both scores on day 14, compared with the respective baseline values, in individual patients.
In the present study, a consistently robust set of criteria was used to prospectively define the inclusion criteria at baseline as PSS and PIS ≥ 2 and define success as a decrease ≥ 1 in PSS and a decrease ≥ 2 in PIS (P = 0.013). On the basis of these criteria, treatment was a failure in 31 of 57 (54.4%) carprofen-treated dogs, compared with 45 of 59 (76.3%) placebo-treated dogs. When a more stringent definition of treatment success (ie, a decrease ≥ 2 in both PSS and PIS) was applied, treatment was classified as a failure in most dogs (39/57 [68.4%] carprofen-treated dogs and 50/59 [84.7%] placebo-treated dogs).
Results of our analysis suggested that the PIS was more sensitive than the PSS in assessing treatment results. This may be because the interference questions are more specific, such as the ability to run, walk, climb, and rise from a recumbent position, whereas the severity score section solicits the owner's estimate of the dog's pain at its worst and least, on average, and the pain “right now.” This is similar to the results of human studies20,21 in which the Brief Pain Inventory was used, and PISs were shown to be more sensitive than PSSs in distinguishing the success of treatment.
Carprofen is used for the control of pain and inflammation associated with osteoarthritis in dogs and was approved by the FDA Center for Veterinary Medicine in 1996. It might therefore seem somewhat surprising that, although the drug has been proven superior to a placebo, treatment was classified as a failure in many carprofen-treated dogs in the present study. However, the results of the study reported here are consistent with results from the pivotal effectiveness studies conducted for the approval of carprofen for this indication and reported through FDA Center for Veterinary Medicine Freedom of Information summaries.15,16 The design of those studies15,16 was similar to that used in the present study. Both veterinarians and owners evaluated dogs before and after 14 days of treatment. In a multisite, randomized, placebo-controlled, blinded study15 of 227 dogs with osteoarthritis (107 that received carprofen [2.2 mg/kg, q 12 h] and 120 given a placebo), veterinarians used a lameness scoring system (on a scale of 1 to 5, with 1 defined as normal) to evaluate lameness, weight bearing, joint mobility, willingness to raise the contralateral limb, and pain. Response to treatment was rated as positive or not positive. Owners and veterinarians rated the same number of caprofen-treated dogs as positive responders (87/107 [81.3%]), although a placebo effect was more evident in owner evaluations, with 30 of 120 (25.0%) of the placebo-treated dogs rated as positive responders by these individuals, compared with 19 of 120 (15.8%) dogs in this group that received the same rating from veterinarians. Cumulative mean lameness scores assigned by veterinarians indicated significantly (P = 0.0006) greater improvement in carprofen-treated dogs (from 12.56 to 9.52), compared with results for placebo-treated dogs (from 12.21 to 11.62).
Another study15 of similar design, conducted in a university setting, also included gait analysis by use of a force plate (n = 70 dogs), and classifying the responses of individual dogs (as positive or not positive) was a better method for evaluating treatment effects than was the use of lameness scores; results were not significantly different between carprofen- and placebo-treated dogs when cumulative lameness scores or the ratings for each of the lameness score components were compared. Although the study investigators were able to distinguish a significant (P = 0.021) treatment effect when rating overall response to treatment (with 18/34 [52.9%] carprofen-treated dogs and 9/36 [25%] placebo-treated dogs classified as having a positive response), lameness scores alone, either cumulative scores or the rating for each of the variables in the lameness score, were insufficient for detection of a treatment effect. Owners ratings were also able to distinguish (P = 0.11) between carprofen and placebo treatment, with 24 of 34 (70.6%) and 14 of 36 (38.9%) dogs, respectively, classified as having a positive response. Results of force plate data analysis showed no significant (P = 0.069) difference between carprofen- and placebo-treated dogs.
A third multicenter study15 with a similar design was conducted to test the effects of carprofen (4.4 mg/kg, q 24 h) in 248 dogs with osteoarthritis. The dog owner and a veterinary investigator were each asked to make a single assessment (composite score) of the severity of the dog's condition before and after treatment, and the veterinarian was also asked to make assessments of lameness and weight bearing, joint mobility, willingness to raise the contralateral limb, and pain on a score of 0 (normal) to 4 (nearly incapacitated). When a positive response was defined as a reduction of at least 1 point in the composite score, positive response rates were 69 of 124 (55.6%) and 53 of 126 (42.1%) for carprofen- and placebo-treated dogs, respectively, on the basis of owner scoring (P = 0.029). Veterinarian assessment yielded positive response rates of 71 of 124 (57.3%) and 52 of 126 (41.3%), respectively, for carprofen- and placebo-treated dogs (P = 0.008) under the same criteria. When a positive response was defined as a composite score reduction ≥ 2, positive response rates on the basis of owner scoring (19/124 [15.3%] and 9/126 [7.1%] for carprofen- and placebo-treated dogs, respectively; P = 0.041) and veterinarian assessment (15/124 [12.1%] and 5/126 [4.0%] for carprofen- and placebo-treated dogs, respectively; P = 0.011) were affected similarly.
In all of these prior studies (described in the FDA Freedom of Information summary),15 consistent with the results of the study reported here, if a large reduction in score (ie, ≥ 2 points) was considered a requirement for treatment success, treatment would be classified as a failure in most dogs, even if owner ratings reflect a benefit of the carprofen treatment, compared with the placebo treatment. It is also interesting to note that in one of these studies,15 the success rate as evaluated by owners for carprofen treatment of dogs with osteoarthritis was 55.6% (with success defined as a reduction of 1 point in the composite score), which is almost identical to the results of the present study when success was defined as a reduction ≥ 1 in both PSS and PIS for population 2 (116 dogs with both PSS and PIS ≥ 2 at baseline on the CBPI), and treatment was classified as successful for 33 of 57 (57.9%) carprofen-treated dogs.
Similar to the 3 previously mentioned studies,15 when the success criteria in the present study were set as a reduction ≥ 1 for both PSS and PIS in population 2, even though treatment was classified as a success in a large number of placebo-treated dogs (37.3%), the difference between treatment groups was significant (P = 0.026). In addition, defining success with more stringent criteria (decrease ≥ 1 in PIS and ≥ 2 in PSS) resulted in treatment being classified as a success in fewer dogs overall, but a substantially smaller P value (P = 0.013) was found. Similar to results of the multicenter study15 that evaluated the effects of carprofen in 248 dogs with osteoarthritis, when success was more strictly defined (in that study15 as a reduction ≥ 2 in the composite score, and in the study reported here as a decrease ≥ 2 in both PSS and PIS), treatment was considered a success in still fewer dogs.
The magnitude of the change in scores detected for dogs in which treatment is considered a success also needs to be clinically relevant if this approach is to be of use to veterinarians when evaluating pain in individual patients and using these changes in scores to help make treatment decisions. One could argue that if owners can notice a difference in their pet's level of pain, as reflected by an improvement in scores on the CPBI, then that improvement is clinically relevant, although it may not be a large difference or indicate that the dog is completely pain free. The present study was not designed to determine clinical relevance of these definitions of treatment success, and studies investigating this aspect are needed. Methods to study clinical relevance generally follow the validation and testing of a subjective outcome assessment instrument. Farrar et al22–25 have examined how to define the clinical importance of a variety of subjective measures of pain in humans. Additional studies that apply the treatment success criteria evaluated in the present study and use the CBPI with different pain interventions will give a robust understanding of the clinical relevancy of the pain scores.
The present study focused on 2 important factors that need to be included in any study to evaluate the effect of a treatment on chronic pain in dogs: the initial inclusion criteria (ie, the minimum pain score specified to allow a dog to enter the study) and the success criteria (ie, how much improvement in the pain score will be required to consider treatment a success). On the basis of pooled data from the present study and a previous study,12 we conclude that a study protocol to evaluate treatment effects in dogs with osteoarthritis will be most useful to detect treatment effects if the inclusion criteria at baseline (day 0) is predefined as a PSS and PIS each ≥ 2 and success for each patient is predefined as a decrease ≥ 1 in PSS and a decrease ≥ 2 in PIS. Although this kind of analysis will require more animals to be enrolled in each arm of a study, compared with an evaluation of median change in scores between groups, it will allow for determination of response at the individual dog level as opposed to the group level, which is key to the pivotal evaluation of intervention efficacy.
It is desirable to be able to design and analyze studies on the basis of a responder analysis (ie, determining a significant difference between groups on the basis of the number of animals that have a positive response to intervention), and this type of study design is required by the FDA Center for Veterinary Medicine for evaluations of drug effectiveness. The results of the present study indicate that the CBPI can be used in the type of protocol design and responder analysis required for regulatory approval of intervention efficacy.
ABBREVIATIONS
| CBPI | Canine Brief Pain Inventory |
| PIS | Pain interference score |
| PSS | Pain severity score |
SAS for Windows, version 9.3, SAS Institute Inc, Cary, NC.
References
1. Wiseman-Orr ML, Nolan AM, Reid J, et al. Development of a questionnaire to measure the effects of chronic pain on health-related quality of life in dogs. Am J Vet Res 2004; 65:1077–1084.
2. Wiseman-Orr ML, Scott EM, Reid J, et al. Validation of a structured questionnaire as an instrument to measure chronic pain in dogs on the basis of effects on health-related quality of life. Am J Vet Res 2006; 67:1826–1836.
3. Hielm-Björkman AK, Rita H, Tulamo RM. Psychometric testing of the Helsinki chronic pain index by completion of a questionnaire in Finnish by owners of dogs with chronic signs of pain caused by osteoarthritis. Am J Vet Res 2009; 70:727–734.
4. Yazbek KV, Fantoni DT. Validity of a health-related quality-of-life scale for dogs with signs of pain secondary to cancer. J Am Vet Med Assoc 2005; 226:1354–1358.
5. University of Pennsylvania Veterinary Clinical Investigation Center. Canine Brief Pain Inventory. Available at: www.CanineBPI.com. Accessed Nov 2, 2012.
6. Brown DC, Boston R, Coyne JC, et al. A novel approach to the use of animals in studies of pain: validation of the Canine Brief Pain Inventory in canine bone cancer. Pain Med 2009; 10:133–142.
7. Brown DC, Boston RC, Coyne JC, et al. Development and psychometric testing of an instrument designed to measure chronic pain in dogs with osteoarthritis. Am J Vet Res 2007; 68:631–637.
8. McDowell I, Newell C. Measuring health: a guide to rating scales and questionnaires. 2nd ed. New York: Oxford University Press, 1996.
9. Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. 3rd ed. New York: Oxford University Press, 2003.
10. Sudman S, Bradburn NM. Asking questions: a practical guide to questionnaire design. San Francisco: Jossey-Bass Inc, 1982.
11. Brown DC, Boston RC, Farrar JT. Comparison of force plate gait analysis and owner assessment of pain using the Canine Brief Pain Inventory in dogs with osteoarthritis. J Vet Intern Med 2013; 27:22–30.
12. Brown DC, Boston RC, Coyne JC, et al. 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.
13. US FDA. Animal and veterinary section. Available at: www.fda.gov/AnimalVeterinary/default.htm. Accessed Nov 2, 2012.
14. Fox SM. Chronic pain in small animal medicine. London: Manson Publishers, 2010.
15. FDA Center for Veterinary Medicine. Freedom of Information summary. Rimadyl caplets (carprofen). NADA No. 141–053. Washington, DC: FDA Center for Veterinary Medicine, 1996.
16. FDA Center for Veterinary Medicine. Freedom of Information summary. Rimadyl chewable tablets. NADA No. 141–111. Washington, DC: FDA Center for Veterinary Medicine, 1996.
17. Moskowitz RW, Sunshine A, Brugger A, et al. American Pain Society pain questionnaire and other pain measures in the assessment of osteoarthritis pain: a pooled analysis of three celecoxib pivotal studies. Am J Ther 2003; 10:12–20.
18. Hochberg MC, Wohlreich M, Gaynor P, et al. Clinically relevant outcomes based on analysis of pooled data from 2 trials of duloxetine in patients with knee osteoarthritis. J Rheumatol 2012; 39:352–358.
19. Jensen MP, Schnitzer TJ, Wang H, et al. Sensitivity of singledomain versus multiple-domain outcome measures to identify responders in chronic low-back pain: pooled analysis of 2 placebo-controlled trials of etoricoxib. Clin J Pain 2012; 28:1–7.
20. Markenson JA, Croft J, Zhang PG, et al. Treatment of persistent pain associated with osteoarthritis with controlled-release oxycodone tablets in a randomized controlled clinical trial. Clin J Pain 2005; 21:524–535.
21. Turk DC, Dworkin RH, Allen RR, et al. Core outcome domains for chronic pain clinical trials: IMMPACT recommendations. Pain 2003; 106:337–45.
22. Farrar JT, Dworkin RH, Max MB. Use of the cumulative proportion of responders analysis graph to present pain data over a range of cut-off points: making clinical trial data more understandable. J Pain Symptom Manage 2006; 31:369–377.
23. Farrar JT, Portenoy RK, Berlin JA, et al. Defining the clinically important difference in pain outcome measures. Pain 2000; 88:287–294.
24. Farrar JT, Pritchett YL, Robinson M, et al. The clinical importance of changes in the 0 to 10 numeric rating scale for worst, least, and average pain intensity: analyses of data from clinical trials of duloxetine in pain disorders. J Pain 2010; 11:109–118.
25. Farrar JT, Young JP Jr, LaMoreaux L, et al. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain 2001; 94:149–158.
