Pain Medicine 2015; 16: 1333–1340 Wiley Periodicals, Inc.
METHODOLOGY, MECHANISMS & TRANSLATIONAL RESEARCH SECTION
Arnold R. Gammaitoni, PharmD,* Jeremiah J. Trudeau, PhD,† Richard Radnovich, DO,‡ Bradley S. Galer, MD,* and Mark P. Jensen, PhD§ *Nuvo Research, Inc, West Chester, Pennsylvania; † Analgesic Solutions, Natick, Massachusetts; ‡Injury Care Medical Center, Boise, Idaho; §Department of Rehabilitation Medicine, University of Washington, Seattle, Washington, USA Arnold R. Gammaitoni and Bradley S. Galer are currently at Zogenix, Inc, San Diego, California, USA. Reprint requests to: Mark P. Jensen, PhD, Department of Rehabilitation Medicine, University of Washington, Box 359612, Harborview Medical Center, 325 Ninth Avenue, Seattle, WA 98104, USA. Voice mail: 206-5433185; Fax: 206-897-4881. E-mail: [email protected]. Disclosures: Dr. Jensen has received grant/research support from and has served as a consultant to Nuvo, and has received consulting fees from Allergan, Covidien, Depomed, Endo Pharmaceuticals, Merck, Pfizer, and RTI Health Solutions, within the past 12 months. Drs. Gammaitoni and Galer were employees of Nuvo when the research was completed. Drs. Jensen, Gammaitoni, and Galer receive royalties from industrysponsored use of the Pain Quality Assessment Scale, but do not receive royalties from non-sponsored use. Dr. Radnovich has been a paid consultant for and has received research funding from Nuvo Research. Funding sources: This study was supported by Nuvo. Author contribution: All authors are responsible for the work described in this paper, and they contributed to
the conception, data interpretation, drafting of the manuscript, and revising the manuscript for intellectual content. All authors provided final approval of the version to be published. Abstract Objectives. No existing pain treatment is effective for all pain problems, and response to pain treatment is highly variable. Knowledge regarding the patient factors that predict response to different treatments could benefit patients by providing an empirical foundation for patient-treatment matching. This study sought to test the hypothesis that improvements following two treatments thought to operate via similar mechanisms would be predicted by similar baseline pain qualities. Design. Prospective prediction analysis using data from a previously published open label trial comparing a heated lidocaine/tetracaine patch versus subacromial corticosteroid injection for the treatment of pain in individuals with shoulder impingement syndrome. Results. Consistent with the study hypothesis, the response to the two treatments were predicted by similar baseline pain qualities; specifically, higher baseline levels of unpleasant, electric, and sensitive pain predicted subsequent improvements in sleep interference, work/activity interference, and patient global ratings of improvement, respectively. Conclusions. The findings are consistent with the combined ideas that (1) those who have the most to gain (i.e., those reporting the highest levels of various pain qualities) can expect the best response to effective treatments and (2) different pain qualities may be associated with different types of outcomes.
1333
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Original Research Articles Predicting Response to Subacromial Injections and Lidocaine/Tetracaine Patch from Pretreatment Pain Quality in Patients with Shoulder Impingement Syndrome
Gammaitoni et al. The findings support further research to examine how pain quality measures may be used to improve patient-treatment matching, and therefore, ultimately improve the efficiency, efficacy, and overall benefit-risk of pain treatment. Key Words. Assessment; Analgesic; Chronic Pain; Steroids Introduction
We recently demonstrated the potential for a baseline measure of pain quality, the Pain Quality Assessment Scale (PQAS [3,4]), for predicting response to pregabalin in a sample of patients with neuropathic pain [5]. We found that fully nine of the 23 PQAS baseline scales and items were significantly associated with response to pregabalin; specifically, the scale scores assessing paroxysmal and deep pain, and the specific items assessing pain intensity, electric, tingling, cramping, radiating, throbbing, and deep pain [5]. Moreover, we found that PQAS items had a sensitively of 85% and specificity of 76% for identifying treatment responders to the active treatment (but not placebo), thereby indicating an ability for PQAS to detect response to specific pharmacology, and not merely response to pain treatment in general. We concluded that additional research was needed to help determine the reliability and generalizability of these findings; for example, if the same pain qualities predicting response to pregabalin also predicted response to other pain treatments and treatment response in other pain populations. The primary aim of this analysis is to address the need for determining the replicability of our initial findings. To address this aim, we used data from an open-label clinical trial comparing a single injection of triamcinolone acetonide with the application of a heated patch containing lidocaine and tetracaine in a sample of patients with shoulder impingement syndrome [6]. Shoulder impingement would be a particularly useful pain condition to study in this context because it can potentially include 1334
Methods Participants and Study Design The data used for this secondary analysis came from a study that compared the effects of a heated lidocaine/ tetracaine (HLT) patch and a subacromial corticosteroid injection on pain intensity in a sample of patients with shoulder impingement syndrome [6]. Following baseline assessment, patients with unilateral shoulder impingement syndrome were randomized to either: 1) a corticosteroid injection into the subscromial space of the painful shoulder or 2) treatment with a HLT patch. Participants assigned to the patch condition were instructed to apply the patch to the painful shoulder for four hours two times every day for 2 weeks at the start of the study. They were then allowed to apply the patch as needed (but not more than twice daily) for next two weeks. No use of the patch was allowed during the last two weeks. We recently reported the findings from another secondary analysis of this same data set which described the effects of the two treatments examined in this study on pain quality profiles [9]. However, the two analyses address different questions; this study examines the ability of baseline pain quality to predict treatment response, while the previous study described the baseline to post-treatment changes in pain quality associated with the two treatments studied [9]. Study Measures Pain Quality We assessed pain quality with the PQAS [3,4]. Each of the 20 PQAS items assess a distinct pain domain or
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Although a great variety of pain treatments are available, no single treatment is effective for all pain problems, and response to pain treatment is highly variable [1,2]. As a result, standard treatment of pain is empirical; most patients require multiple trials with a number of different treatments both alone and in combination, before a regimen is identified that provides the greatest benefit with the best tolerability for each individual patient. Each new trial takes time, often many weeks, during which patients may experience prolonged significant pain, functional impairment, and potential side effects from treatment. The availability of strategies that could be used to identify patients who are most likely to be treatment responders early in the course of therapy or even before selecting the initial treatment would be of significant benefit to clinicians and patients alike, saving both time and improving the benefit-risk profile of the treatment(s) selected for the individual patient.
nociceptive, neuropathic, or mixed elements, and as such each patient may report different pain qualities and respond differently to treatments. Both the treatments examined in this study are hypothesized to reduce pain via similar mechanisms; that is, by reducing ectopic signal generation by blocking sodium channels [7]. Therefore, we did not anticipate significant differences between the two treatments in the factors that predicted treatment response. However, we were particularly interested in determining if the same (or similar) factors that predicted response to pregabalin also predicted response to the triamcinolone acetonide injection and application of a heated patch containing lidocaine and tetracaine. In addition, we extended the analyses to examine more than only treatment response as defined as a reduction in average pain intensity [8]; specifically, we expanded the analyses to also include: 1) change in worst pain intensity; 2) change in sleep interference; 3) change in activity/work interference; and 4) post-treatment patient global assessment of change. As research in this area is new, we did not have specific a priori hypotheses regarding which baseline pain quality measures would or would not predict treatment response. Thus, it represents a preliminary hypothesis generating study to help determine if continued research along these lines is warranted.
Predicting Treatment Response quality. Fifteen of the items can be combined to create three scale scores assessing Paroxysmal, Surface, and Deep pain [10]. The PQAS items can also be examined as individual ratings. The participants were administered the PQAS at baseline (Day 1), and at the end of each 2week phase (i.e., Days 14, 28, and 42). However, only the baseline and post-treatment (day 42) data were used for this analyses. Average and Worst Pain Intensity
Sleep, Activity, and Work Interference How much pain interfered with subjects general activity, normal work, and sleep over the previous 24 hours were all assessed using 0–10 numerical rating scales where 0 was “Did not interfere” and 10 was “Completely interfered.” The questions were asked in the format “How much did the pain in your shoulder interfere with your during the past 24 hours?” Numerical rating scales of pain interference have a great deal of evidence supporting their validity and are commonly used in pain research [10,12]. Because activity and work item scores were very strongly associated with one another (r5 0.85, P < 0.001), we combined them into a single composite activity/work score by computing the mean of the two items. Patient Global Assessment of Change At post-treatment (day 42), study participants were asked to complete a 7-point Patient Global Impression of Change (PGIC) rating, ranging from 1 (Very much worse) to 7 (Very much improved). The patient global evaluation of outcome is noted as an important outcome domain by consensus groups [12,13], and the categorical PGIC measures such as the one used in this study are commonly used to assess this domain in pain clinical trials [14,15]. Data Analysis To examine the zero-order associations between the PQAS scales and item scores and outcome, we com-
Results Participants Sixty participants were enrolled in this study. Forty-six (23 per treatment condition) completed the study [6]. The demographic and baseline pain quality information from the participants who provided complete data can be seen in Table 1. Univariate Associations Between Baseline PQAS Scales and Items and Outcome Pre- to post-treatment improvement in average pain correlated significantly with the individual PQAS items for hot pain quality (r 5 20.29, P < 0.05) and aching pain quality (r 5 20.29, P < 0.05); that is when pain was described as more hot or more achy at baseline, participants tended to report less improvement (reduction) in global pain intensity with treatment. No baseline PQAS item or subscale scores correlated significantly with change in the worst pain intensity ratings. Pre- to post-treatment improvement in sleep interference correlated significantly with baseline ratings of pain unpleasantness (r 5 0.34, P < 0.05); higher baseline pain unpleasantness was associated with more improvement (a greater reduction in sleep interference). Improvement in the activity/work composite score correlated significantly with both electric quality (r 5 0.39, P < 0.01) and the surface intensity item (r 5 0.30, P < 0.05); higher baseline levels of these pain qualities were associated with greater improvement (reduction in interference with activity/work). Finally, baseline ratings of the sensitive pain item correlated significantly 1335
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Recalled average and worst pain intensity in the past 24 hours were assessed using 0–10 numerical rating scales with 0 5 “No pain” and 10 5 “Pain as bad as you can imagine” on daily dairies throughout their participation in the study. A great deal of evidence supports the reliability and validity of 0–10 numerical scales for assessing pain intensity [11]. Moreover, 0–10 measures are judged to have the most strengths and fewest weaknesses of all available pain intensity measures, and so are recommended by expert consensus groups as the measures of choice for pain clinical trials [12,13]. Post-treatment scores for both these variables were then subtracted from the pre-treatment scores to represent treatmentrelated change in both average and worst pain intensity.
puted Pearson correlations between these variables for both treatment groups combined as well as for the treatment groups individually. We then used linear regression analyses to determine the ability of the baseline PQAS items to predict treatment outcome. Given the large number of potential predictors (20 PQAS items) and the small number of participants in this study, we selected only those items that showed a significant (P < 0.05) univariate association with an outcome variable as potential predictors in the regression analyses. Regression analyses were conducted in a stepped fashion. In the first step, the model included only the pain quality baseline scores that correlated significantly with the outcome variable. In the second step, treatment condition (injection versus patch) was added as a potential predictor variable and in the third step interaction terms between treatment and pain quality scores were added. Significant relationships were plotted for descriptive purposes. No adjustment for multiple statistical testing was performed as the analyses were post hoc and exploratory. Additionally, to address possible issues of colinearity, the models were repeated after centering the predictor variables; no meaningful differences were found compared with the raw (uncentered) versions.
Gammaitoni et al.
Table 1 Demographic and baseline pain quality information about the study participants Treatment Condition
HLT Patch
Subacromial Injection
The final model predicting improvement in average pain intensity from the PQAS hot and aching items was significant (P < 0.05) with an R2 of 0.26. Interestingly, however, the only element with a significant coefficient within the entire model was the interaction term for hot pain by treatment condition (beta 5 20.56, P < 0.05). Follow-up regressions were conducted on each treatment arm separately to help explain this interaction, using the hot and aching pain at baseline as predictors. When only patch-treated subjects were examined the model was nonsignificant overall (R2 5 0.09, P 5 0.38). However, when injection-treated subjects were examined the model was significant (R2 5 0.37, P < 0.05) and the hot pain quality item had a significant coefficient (Beta 5 20.53, P < 0.01). This relationship indicated that subjects with lower scores on the hot pain quality item of the PQAS and in the injection treatment arm were more likely to experience a greater improvement in their overall average pain intensity than subjects with higher baseline hot pain ratings. This relationship is illustrated in Figure 1. As can be seen, this finding is related to the fact that participants reporting no hot pain at baseline evidenced a large range of response (from a pain reduction of only 1 point to a pain reduction of 8 points on a 0–10 scale). Those with higher than 0 hot pain ratings reported a lower range in treatment response (i.e., reductions from 1 to 5 points).
Improvement in Sleep Interference The first step model predicting change in sleep interference from unpleasant pain quality was significant (R 25 0.12, P < 0.05) and unpleasant pain quality (necessarily as the only predictor) had a significant coefficient (Beta 5 0.34, P < 0.05). However, with the second and third steps adding treatment condition and the Unpleasant Pain X Treatment Condition interaction model became only marginally significant (R2 change 5 0.15, P 5 0.08 at step 3) and none of the individual model
(r 5 0.32, P < 0.05) with the PGIC rating, with higher ratings on this item associated with higher ratings of global improvement. These PQAS items were therefore selected as the potential predictors for each respective outcome in the subsequent regression models. No additional variables were identified as significant predictors of outcome when examined for each treatment condition separately. Regression Analyses Predicting Outcome from Baseline PQAS Items Given that no baseline pain qualities were associated significantly with improvement in 24-hour worst pain, no regression analyses were conducted using that outcome. 1336
Figure 1 Association between reduction in global pain intensity and baseline “hot” pain rating (injection group only, n 5 23).
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
N 23 23 Age Mean (SD) 50.5 (12.4) 51.5 (10.0) Range 28–72 18–68 Sex Men, N (%) 14 (61%) 17 (74%) Women, N (%) 9 (39%) 6 (26%) PQAS pain quality scale scores at baseline Deep 4.2 (2.0) 4.2 (1.5) Paroxysmal 4.1 (2.1) 4.5 (1.5) Surface 1.5 (1.9) 1.7 (1.7) PQAS pain quality item scores at baseline Intense 6.7 (1.5) 6.3 (1.5) Sharp 6.2 (2.7) 6.7 (2.3) Hot 2.0 (2.7) 2.3 (2.4) Dull 4.7 (3.1) 4.7 (2.7) Cold 0.5 (1.6) 1.0 (1.9) Sensitive 1.4 (2.7) 1.2 (1.7) Tender 5.2 (3.1) 4.3 (2.1) Itchy 0.6 (1.6) 0.9 (1.2) Shooting 5.7 (2.8) 6.1 (1.9) Numb 2.7 (3.1) 2.8 (3.2) Electrical 2.6 (3.1) 3.5 (3.1) Tingling 2.3 (3.2) 2.5 (3.0) Cramping 2.6 (3.5) 2.7 (2.6) Radiating 4.0 (3.4) 4.0 (2.5) Throbbing 4.5 (3.4) 4.0 (2.7) Aching 5.4 (2.4) 5.8 (2.1) Heavy 2.6 (3.2) 3.6 (2.7) Unpleasant 6.6 (1.8) 6.7 (1.6) Deep 6.7 (1.9) 6.8 (1.5) Surface 2.3 (2.8) 2.6 (2.0)
Improvement in Average Pain Intensity
Predicting Treatment Response ciated with a greater range of improvement from reports of worsening of activity/work interference by 6 points to improvement by 7.5 points), and higher baseline electric pain quality scores associated with only improvement ratings, ranging from 4 to 9 points for those whose baseline electric pain ratings were 7 or higher.
Patient Global Impression of Change
components retained a significant coefficient. We therefore considered primarily the significant model, which showed subjects with higher unpleasantness quality at baseline experiencing greater improvement in sleep interference (more reduced interference score) irrespective of treatment. This finding is illustrated in Figure 2. These data indicate that most participants found their pain to be unpleasant, with baseline scores for all but three subjects ranging from 5 to 10 on a 0 to 10 scale. Moreover, although most of the subjects reported decrease in sleep interference with treatment, those whose baseline unpleasantness score of 7 or greater evidenced a larger range in improvement (redetections ranging from 0 to 10 points), whereas those whose baseline unpleasantness ratings were 5 and 6 showed a more restricted range in improvement (reductions ranging from 0 to 7.5).
Improvement in Activity/Work Interference A similar pattern is seen when predicting interference with work and activities using electric and surface intensity pain qualities. The model is significant at the first step (R2 5 0.16, P < 0.05) with electric pain quality displaying a marginally significant coefficient weight (Beta 5 0.32, P 5 0.06) but surface intensity not being a significant predictor (P 5 0.47). The second step model showed similar results; the model was significant overall (R2 5 0.18, P < 0.05) with a nearly significant coefficient for electric quality (Beta 5 0.34, P 5 0.05) but nonsignificant coefficients for surface intensity or treatment condition (Beta 5 0.12 and 20.12, P 5 0.49 and 0.41, respectively). The model ceases to be significant when the interaction terms are added (overall R2 5 0.18, P 5 0.14) and no variable coefficients are significant when all of the variables are entered. The relationship of higher baseline electric pain quality scores to greater improvement in activity/work interference (more reduced interference score) is illustrated in Figure 3. As can be seen, low baseline electric pain quality scores are asso-
Discussion The study findings provide additional support for the ability of baseline pain quality measures to predict pain reduction following treatment, but extend the findings to include additional outcome measures, and indicate that different pain qualities may be associated with different outcomes. These results contribute to the growing body of research supporting the importance of examining pain qualities (in addition to pain intensity) in clinical and
Figure 3 Association between reduction in activity/work interference and baseline electric pain rating. 1337
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Figure 2 Association between reduction in sleep interference and baseline unpleasantness pain rating.
The model predicting PGIC scores from sensitive pain quality were similar to the pattern of results seen in predicting sleep and activity/work interference. PGIC assessed at post-treatment was significantly associated with the baseline sensitivity item (R2 5 0.11, P < 0.05), but subsequent steps adding treatment and the Treatment X Sensitivity ratings interaction resulted in a nonsignificant final model (P 5 0.14). At the first step the sole predictor, sensitive pain quality, had a coefficient beta of 0.32 (P < 0.05) showing subjects with higher scores in sensitive pain at baseline being more likely to report greater overall impressions of change (illustrated shown in see Figure 4). As can be seen, the pattern for findings for baseline sensitive pain predicting subsequent global ratings are much like the pattern found for electric pain predicting a decrease in activity/work interference; those with lower baseline sensitive scores evidence a larger range of outcomes, while those with higher baseline sensitive scores show both a reduced range in outcome, and an overall better response.
Gammaitoni et al.
research settings, and suggest those pain qualities that might be most important—at least in individuals with shoulder impingement pain. Our hypothesis that the pain qualities predicting treatment response would be similar across the two treatments examined (due to their similar mechanisms) was largely supported; in both treatment conditions, higher baseline levels of unpleasant pain, electric pain, and sensitive pain predicted a priori improvements in sleep interference, work/activity interference, and patient global ratings of improvement, respectively. In general, across these baseline measures, it was the participants who reported severe levels of these pain qualities (7 or higher on a 0–10 scale) that evidenced the most consistent improvement in outcome. These findings are consistent with the combined ideas that 1) those who have the most to gain (i.e., those reporting the highest baseline levels of various pain qualities) can expect the best response to effective treatments and 2) different pain qualities may be associated with different types of outcomes; specifically that pain unpleasantness may be more closely linked to sleep interference, electric pain sensations may be more closely linked to activity/work interference, and that sensitive pain may be most closely linked to global evaluations of the pain experience. In a previous study, we found that pain unpleasantness was among the strongest consistent predictors of sleep interference in three different samples of individuals with chronic pain [16]. The current findings are consistent with our earlier findings, and suggest that when pain is rated as severely unpleasant, even when or if it is not rated as severely intense, clinicians might pay particular attention to pain-related sleep problems as well. Given the importance of sleep to global health in general [17] as well as to pain management in particular [18], these patients may benefit from consideration of treatments for sleep problems, as appropriate. 1338
One pain quality was not a consistent predictor of treatment outcome across both interventions—hot pain. Moreover, the direction of associations found for this pain quality as a predictor of outcome differed from those found for the other predictors. Specifically, less hot pain at baseline for the participants who received the injection was (strongly) associated with more pain intensity reduction among the participants who received the injection only. Baseline levels of hot pain were only weakly (and not significantly) associated with treatment response to the patch. Again, given the fact that this study is the first time these associations have been examined for these treatments, it remains possible that the finding will not replicate. But the finding does suggest the possibility that despite similar hypothesized mechanisms of these treatment (i.e., blockade of sodium channels), the delivery method (injection versus patch) may influence outcome. Those with high hot pain, for example, might not benefit from the injection, whereas those with low hot pain could potentially benefit more from the injection. At this point, we are not recommending that treatment decisions be made based on the current findings alone; we are suggesting, however, that the findings strongly support the need for more research in this area to help determine which of the current findings replicate. The study has some important limitations that should be considered when interpreting the results. First, as already discussed, research examining the importance of baseline pain quality as a predictor of pain treatment outcome is still in its infancy. More studies evaluating a variety of treatments in a variety of pain populations are needed to help sort out which findings are most reliable. Only when multiple studies have been completed, will we be able to make empirically guided recommendations regarding the use of pain quality measures for guiding treatment. Related to this issue, there were a number of pain qualities that were relatively low at baseline. These included, for example, sharp, shooting, hot, and electric pain. Low baseline levels reduce the potential of variance due to floor effects, which can attenuate true associations that exist in the population. It is interesting, of course that two of these qualities, hot and electric, still emerged as significant predictors of
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Figure 4 Association between patient global ratings of improvement and baseline sensitive pain rating.
To our knowledge, this study is the first that has examined the ability of baseline pain measures to predict treatment-related reductions in work and activity interference and patient global ratings of improvement. Thus, replication is needed before we can conclude that electric pain sensations and sensitive pain are specifically important to these outcomes. However, if replicated in samples from additional chronic pain populations, the findings would indicate particular importance for these pain qualities and domains, and perhaps a particular focus on identifying effective treatments for electric and sensitive pain, at least in order for treatment to also result in meaningful reductions of the extent to which pain interferes with function and the patient’s overall evaluation of treatment success.
Predicting Treatment Response
Despite the study limitations, however, the results provide further support for the potential of baseline pain qualities to predict who will response to different pain treatments (and how much). As such, they may ultimately provide an important clinical utility for helping guide treatment decisions if findings are replicated. The findings provide support for investigating this potential further.
Acknowledgments This study was supported by Nuvo. All authors are responsible for the work described in this paper, and they contributed to the conception, data interpretation, drafting of the manuscript, and revising the manuscript for intellectual content. All authors provided final approval of the version to be published.
References 1 Preston SJ, Arnold MH, Beller EM, Brooks PM, Buchanan WW. Variability in response to nonsteroidal anti-inflammatory analgesics: Evidence from controlled clinical therapeutic trial of flurbiprofen in rheumatoid arthritis. Br J Clin Pharmacol 1988;26:759–64. 2 Turk DC, Wilson HD, Cahana A. Treatment of chronic non-cancer pain. Lancet 2011;377:2226– 35. 3 Jensen MP, Galer BS, Gammaitoni AR, Lin CP. The Pain Quality Assessment Scale (PQAS) and Revised Pain Quality Assessment Scale (PQAS-R): Manual and user Guide, 2010. Available at: [email protected]. Last accessed March 18, 2015. 4 Jensen MP, Gammaitoni AR, Olaleye DO, Oleka N, Nalamachu SR, Galer BS. The pain quality assess-
ment scale: assessment of pain quality in carpal tunnel syndrome. J Pain 2006;7:823–32. 5 Gammaitoni AR, Smugar SS, Jensen MP, et al. Predicting response to pregabalin from pretreatment pain quality: Clinical applications of the pain quality assessment scale. Pain Med 2012;14:526– 32. 6 Radnovich R, Trudeau J, Gammaitoni A. A randomized clinical study of the heated lidocaine/ tetracaine patch versus subacromial corticosterioid injection for the treatment of pain assocaited with shoulder impingement syndrome. J Pain Res 2014; 7:727–35. 7 Watson CPN. Topical local anesthestics for neuropathic pain. In: Hansson PT, Fields HL, Hill RG, Marchettini P, editors. Neuropathic pain: Pathophysiology and treatment. Seattle: IASP Press; 2001: 215–21. 8 Turk DC, Dworkin RH, Allen RR, et al. Core outcome domains for chronic pain clinical trials: IMMPACT recommendations. Pain 2003;106:337– 45. 9 Jensen MP, Trudeau JJ, Radnovich R, Galer BS, Gammaitoni AR. The pain quality response profile of corticosteriod injections and heated lidocaine/tetracaine patch in the treatment of shoulder impingement syndrome. Clin J Pain, in press. 10 Victor TW, Jensen M, Gammaitoni AR, Gould EM, White RE, Galer BS. The dimensions of pain quality: Factor analysis of the Pain Quality Assessment Scale. Clin J Pain 2008;24:550–5. 11 Jensen MP, Karoly P. Self-report scales and procedures for assessing pain in adults. In: Turk DC, Melzack R, editors. Handbook of pain assessmenta New York: Guilford Press; 2011:19–44. 12 Turk DC, Dworkin RH, McDermott MP, et al. Analyzing multiple endpoints in clinical trials of pain treatments: IMMPACT recommendations. Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials. Pain 2008;139:485–93. 13 Dworkin RH, Turk DC, Farrar JT, et al. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. Pain 2005;113:9– 19. 14 Hosomi K, Shimokawa T, Ikoma K, et al. Daily repetitive transcranial magnetic stimulation of primary motor cortex for neuropathic pain: A randomized, multicenter, double-blind, crossover, shamcontrolled trial. Pain 2013;154:1065–72.
1339
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
treatment outcome despite their low baseline levels. This provides some support for the potential reliability of the findings related to these two pain qualities. It is also possible, even likely, that different pain qualities would predict outcome if different outcome measures were used. This idea is supported in part by our finding that different pain qualities predicted improvement in the four different outcomes measures examined here. Additional research would be needed to determine how pain assessed at baseline predict response not just to different treatments but as measured by different outcome variables. Finally, as noted previously, this analyses involved a large number of statistical tests without control for alpha inflation. Because of this, the findings should be considered exploratory and not confirmatory. Thus, additional research using larger sample sizes if possible and in also in samples of patients with more variability in baseline pain quality ratings would help to determine the reliability of these findings and may help identify additional predictors or allow for even stronger associations than those identified here.
Gammaitoni et al. 15 Rowbotham M, Harden N, Stacey B, Bernstein P, Magnus-Miller L. Gabapentin for the treatment of postherpetic neuralgia: A randomized controlled trial. JAMA 1998;280:1837–42. 16 Jensen MP, Dworkin RH, Gammaitoni AR, Olaleye DO, Oleka N, Galer BS. Do pain qualities and spatial characteristics make independent contributions to
interference with physical and emotional functioning? J Pain 2006;7:644–53. 17 Wells ME, Vaughn BV. Poor sleep challenging the health of a Nation. Neurodiagn J 2012;52:233–49. 18 Okifuji A, Hare BD. Do sleep disorders contribute to pain sensitivity? Curr Rheumatol Rep 2011;13:528–34.
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
1340
METHODOLOGY, MECHANISMS & TRANSLATIONAL RESEARCH SECTION
Arnold R. Gammaitoni, PharmD,* Jeremiah J. Trudeau, PhD,† Richard Radnovich, DO,‡ Bradley S. Galer, MD,* and Mark P. Jensen, PhD§ *Nuvo Research, Inc, West Chester, Pennsylvania; † Analgesic Solutions, Natick, Massachusetts; ‡Injury Care Medical Center, Boise, Idaho; §Department of Rehabilitation Medicine, University of Washington, Seattle, Washington, USA Arnold R. Gammaitoni and Bradley S. Galer are currently at Zogenix, Inc, San Diego, California, USA. Reprint requests to: Mark P. Jensen, PhD, Department of Rehabilitation Medicine, University of Washington, Box 359612, Harborview Medical Center, 325 Ninth Avenue, Seattle, WA 98104, USA. Voice mail: 206-5433185; Fax: 206-897-4881. E-mail: [email protected]. Disclosures: Dr. Jensen has received grant/research support from and has served as a consultant to Nuvo, and has received consulting fees from Allergan, Covidien, Depomed, Endo Pharmaceuticals, Merck, Pfizer, and RTI Health Solutions, within the past 12 months. Drs. Gammaitoni and Galer were employees of Nuvo when the research was completed. Drs. Jensen, Gammaitoni, and Galer receive royalties from industrysponsored use of the Pain Quality Assessment Scale, but do not receive royalties from non-sponsored use. Dr. Radnovich has been a paid consultant for and has received research funding from Nuvo Research. Funding sources: This study was supported by Nuvo. Author contribution: All authors are responsible for the work described in this paper, and they contributed to
the conception, data interpretation, drafting of the manuscript, and revising the manuscript for intellectual content. All authors provided final approval of the version to be published. Abstract Objectives. No existing pain treatment is effective for all pain problems, and response to pain treatment is highly variable. Knowledge regarding the patient factors that predict response to different treatments could benefit patients by providing an empirical foundation for patient-treatment matching. This study sought to test the hypothesis that improvements following two treatments thought to operate via similar mechanisms would be predicted by similar baseline pain qualities. Design. Prospective prediction analysis using data from a previously published open label trial comparing a heated lidocaine/tetracaine patch versus subacromial corticosteroid injection for the treatment of pain in individuals with shoulder impingement syndrome. Results. Consistent with the study hypothesis, the response to the two treatments were predicted by similar baseline pain qualities; specifically, higher baseline levels of unpleasant, electric, and sensitive pain predicted subsequent improvements in sleep interference, work/activity interference, and patient global ratings of improvement, respectively. Conclusions. The findings are consistent with the combined ideas that (1) those who have the most to gain (i.e., those reporting the highest levels of various pain qualities) can expect the best response to effective treatments and (2) different pain qualities may be associated with different types of outcomes.
1333
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Original Research Articles Predicting Response to Subacromial Injections and Lidocaine/Tetracaine Patch from Pretreatment Pain Quality in Patients with Shoulder Impingement Syndrome
Gammaitoni et al. The findings support further research to examine how pain quality measures may be used to improve patient-treatment matching, and therefore, ultimately improve the efficiency, efficacy, and overall benefit-risk of pain treatment. Key Words. Assessment; Analgesic; Chronic Pain; Steroids Introduction
We recently demonstrated the potential for a baseline measure of pain quality, the Pain Quality Assessment Scale (PQAS [3,4]), for predicting response to pregabalin in a sample of patients with neuropathic pain [5]. We found that fully nine of the 23 PQAS baseline scales and items were significantly associated with response to pregabalin; specifically, the scale scores assessing paroxysmal and deep pain, and the specific items assessing pain intensity, electric, tingling, cramping, radiating, throbbing, and deep pain [5]. Moreover, we found that PQAS items had a sensitively of 85% and specificity of 76% for identifying treatment responders to the active treatment (but not placebo), thereby indicating an ability for PQAS to detect response to specific pharmacology, and not merely response to pain treatment in general. We concluded that additional research was needed to help determine the reliability and generalizability of these findings; for example, if the same pain qualities predicting response to pregabalin also predicted response to other pain treatments and treatment response in other pain populations. The primary aim of this analysis is to address the need for determining the replicability of our initial findings. To address this aim, we used data from an open-label clinical trial comparing a single injection of triamcinolone acetonide with the application of a heated patch containing lidocaine and tetracaine in a sample of patients with shoulder impingement syndrome [6]. Shoulder impingement would be a particularly useful pain condition to study in this context because it can potentially include 1334
Methods Participants and Study Design The data used for this secondary analysis came from a study that compared the effects of a heated lidocaine/ tetracaine (HLT) patch and a subacromial corticosteroid injection on pain intensity in a sample of patients with shoulder impingement syndrome [6]. Following baseline assessment, patients with unilateral shoulder impingement syndrome were randomized to either: 1) a corticosteroid injection into the subscromial space of the painful shoulder or 2) treatment with a HLT patch. Participants assigned to the patch condition were instructed to apply the patch to the painful shoulder for four hours two times every day for 2 weeks at the start of the study. They were then allowed to apply the patch as needed (but not more than twice daily) for next two weeks. No use of the patch was allowed during the last two weeks. We recently reported the findings from another secondary analysis of this same data set which described the effects of the two treatments examined in this study on pain quality profiles [9]. However, the two analyses address different questions; this study examines the ability of baseline pain quality to predict treatment response, while the previous study described the baseline to post-treatment changes in pain quality associated with the two treatments studied [9]. Study Measures Pain Quality We assessed pain quality with the PQAS [3,4]. Each of the 20 PQAS items assess a distinct pain domain or
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Although a great variety of pain treatments are available, no single treatment is effective for all pain problems, and response to pain treatment is highly variable [1,2]. As a result, standard treatment of pain is empirical; most patients require multiple trials with a number of different treatments both alone and in combination, before a regimen is identified that provides the greatest benefit with the best tolerability for each individual patient. Each new trial takes time, often many weeks, during which patients may experience prolonged significant pain, functional impairment, and potential side effects from treatment. The availability of strategies that could be used to identify patients who are most likely to be treatment responders early in the course of therapy or even before selecting the initial treatment would be of significant benefit to clinicians and patients alike, saving both time and improving the benefit-risk profile of the treatment(s) selected for the individual patient.
nociceptive, neuropathic, or mixed elements, and as such each patient may report different pain qualities and respond differently to treatments. Both the treatments examined in this study are hypothesized to reduce pain via similar mechanisms; that is, by reducing ectopic signal generation by blocking sodium channels [7]. Therefore, we did not anticipate significant differences between the two treatments in the factors that predicted treatment response. However, we were particularly interested in determining if the same (or similar) factors that predicted response to pregabalin also predicted response to the triamcinolone acetonide injection and application of a heated patch containing lidocaine and tetracaine. In addition, we extended the analyses to examine more than only treatment response as defined as a reduction in average pain intensity [8]; specifically, we expanded the analyses to also include: 1) change in worst pain intensity; 2) change in sleep interference; 3) change in activity/work interference; and 4) post-treatment patient global assessment of change. As research in this area is new, we did not have specific a priori hypotheses regarding which baseline pain quality measures would or would not predict treatment response. Thus, it represents a preliminary hypothesis generating study to help determine if continued research along these lines is warranted.
Predicting Treatment Response quality. Fifteen of the items can be combined to create three scale scores assessing Paroxysmal, Surface, and Deep pain [10]. The PQAS items can also be examined as individual ratings. The participants were administered the PQAS at baseline (Day 1), and at the end of each 2week phase (i.e., Days 14, 28, and 42). However, only the baseline and post-treatment (day 42) data were used for this analyses. Average and Worst Pain Intensity
Sleep, Activity, and Work Interference How much pain interfered with subjects general activity, normal work, and sleep over the previous 24 hours were all assessed using 0–10 numerical rating scales where 0 was “Did not interfere” and 10 was “Completely interfered.” The questions were asked in the format “How much did the pain in your shoulder interfere with your during the past 24 hours?” Numerical rating scales of pain interference have a great deal of evidence supporting their validity and are commonly used in pain research [10,12]. Because activity and work item scores were very strongly associated with one another (r5 0.85, P < 0.001), we combined them into a single composite activity/work score by computing the mean of the two items. Patient Global Assessment of Change At post-treatment (day 42), study participants were asked to complete a 7-point Patient Global Impression of Change (PGIC) rating, ranging from 1 (Very much worse) to 7 (Very much improved). The patient global evaluation of outcome is noted as an important outcome domain by consensus groups [12,13], and the categorical PGIC measures such as the one used in this study are commonly used to assess this domain in pain clinical trials [14,15]. Data Analysis To examine the zero-order associations between the PQAS scales and item scores and outcome, we com-
Results Participants Sixty participants were enrolled in this study. Forty-six (23 per treatment condition) completed the study [6]. The demographic and baseline pain quality information from the participants who provided complete data can be seen in Table 1. Univariate Associations Between Baseline PQAS Scales and Items and Outcome Pre- to post-treatment improvement in average pain correlated significantly with the individual PQAS items for hot pain quality (r 5 20.29, P < 0.05) and aching pain quality (r 5 20.29, P < 0.05); that is when pain was described as more hot or more achy at baseline, participants tended to report less improvement (reduction) in global pain intensity with treatment. No baseline PQAS item or subscale scores correlated significantly with change in the worst pain intensity ratings. Pre- to post-treatment improvement in sleep interference correlated significantly with baseline ratings of pain unpleasantness (r 5 0.34, P < 0.05); higher baseline pain unpleasantness was associated with more improvement (a greater reduction in sleep interference). Improvement in the activity/work composite score correlated significantly with both electric quality (r 5 0.39, P < 0.01) and the surface intensity item (r 5 0.30, P < 0.05); higher baseline levels of these pain qualities were associated with greater improvement (reduction in interference with activity/work). Finally, baseline ratings of the sensitive pain item correlated significantly 1335
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Recalled average and worst pain intensity in the past 24 hours were assessed using 0–10 numerical rating scales with 0 5 “No pain” and 10 5 “Pain as bad as you can imagine” on daily dairies throughout their participation in the study. A great deal of evidence supports the reliability and validity of 0–10 numerical scales for assessing pain intensity [11]. Moreover, 0–10 measures are judged to have the most strengths and fewest weaknesses of all available pain intensity measures, and so are recommended by expert consensus groups as the measures of choice for pain clinical trials [12,13]. Post-treatment scores for both these variables were then subtracted from the pre-treatment scores to represent treatmentrelated change in both average and worst pain intensity.
puted Pearson correlations between these variables for both treatment groups combined as well as for the treatment groups individually. We then used linear regression analyses to determine the ability of the baseline PQAS items to predict treatment outcome. Given the large number of potential predictors (20 PQAS items) and the small number of participants in this study, we selected only those items that showed a significant (P < 0.05) univariate association with an outcome variable as potential predictors in the regression analyses. Regression analyses were conducted in a stepped fashion. In the first step, the model included only the pain quality baseline scores that correlated significantly with the outcome variable. In the second step, treatment condition (injection versus patch) was added as a potential predictor variable and in the third step interaction terms between treatment and pain quality scores were added. Significant relationships were plotted for descriptive purposes. No adjustment for multiple statistical testing was performed as the analyses were post hoc and exploratory. Additionally, to address possible issues of colinearity, the models were repeated after centering the predictor variables; no meaningful differences were found compared with the raw (uncentered) versions.
Gammaitoni et al.
Table 1 Demographic and baseline pain quality information about the study participants Treatment Condition
HLT Patch
Subacromial Injection
The final model predicting improvement in average pain intensity from the PQAS hot and aching items was significant (P < 0.05) with an R2 of 0.26. Interestingly, however, the only element with a significant coefficient within the entire model was the interaction term for hot pain by treatment condition (beta 5 20.56, P < 0.05). Follow-up regressions were conducted on each treatment arm separately to help explain this interaction, using the hot and aching pain at baseline as predictors. When only patch-treated subjects were examined the model was nonsignificant overall (R2 5 0.09, P 5 0.38). However, when injection-treated subjects were examined the model was significant (R2 5 0.37, P < 0.05) and the hot pain quality item had a significant coefficient (Beta 5 20.53, P < 0.01). This relationship indicated that subjects with lower scores on the hot pain quality item of the PQAS and in the injection treatment arm were more likely to experience a greater improvement in their overall average pain intensity than subjects with higher baseline hot pain ratings. This relationship is illustrated in Figure 1. As can be seen, this finding is related to the fact that participants reporting no hot pain at baseline evidenced a large range of response (from a pain reduction of only 1 point to a pain reduction of 8 points on a 0–10 scale). Those with higher than 0 hot pain ratings reported a lower range in treatment response (i.e., reductions from 1 to 5 points).
Improvement in Sleep Interference The first step model predicting change in sleep interference from unpleasant pain quality was significant (R 25 0.12, P < 0.05) and unpleasant pain quality (necessarily as the only predictor) had a significant coefficient (Beta 5 0.34, P < 0.05). However, with the second and third steps adding treatment condition and the Unpleasant Pain X Treatment Condition interaction model became only marginally significant (R2 change 5 0.15, P 5 0.08 at step 3) and none of the individual model
(r 5 0.32, P < 0.05) with the PGIC rating, with higher ratings on this item associated with higher ratings of global improvement. These PQAS items were therefore selected as the potential predictors for each respective outcome in the subsequent regression models. No additional variables were identified as significant predictors of outcome when examined for each treatment condition separately. Regression Analyses Predicting Outcome from Baseline PQAS Items Given that no baseline pain qualities were associated significantly with improvement in 24-hour worst pain, no regression analyses were conducted using that outcome. 1336
Figure 1 Association between reduction in global pain intensity and baseline “hot” pain rating (injection group only, n 5 23).
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
N 23 23 Age Mean (SD) 50.5 (12.4) 51.5 (10.0) Range 28–72 18–68 Sex Men, N (%) 14 (61%) 17 (74%) Women, N (%) 9 (39%) 6 (26%) PQAS pain quality scale scores at baseline Deep 4.2 (2.0) 4.2 (1.5) Paroxysmal 4.1 (2.1) 4.5 (1.5) Surface 1.5 (1.9) 1.7 (1.7) PQAS pain quality item scores at baseline Intense 6.7 (1.5) 6.3 (1.5) Sharp 6.2 (2.7) 6.7 (2.3) Hot 2.0 (2.7) 2.3 (2.4) Dull 4.7 (3.1) 4.7 (2.7) Cold 0.5 (1.6) 1.0 (1.9) Sensitive 1.4 (2.7) 1.2 (1.7) Tender 5.2 (3.1) 4.3 (2.1) Itchy 0.6 (1.6) 0.9 (1.2) Shooting 5.7 (2.8) 6.1 (1.9) Numb 2.7 (3.1) 2.8 (3.2) Electrical 2.6 (3.1) 3.5 (3.1) Tingling 2.3 (3.2) 2.5 (3.0) Cramping 2.6 (3.5) 2.7 (2.6) Radiating 4.0 (3.4) 4.0 (2.5) Throbbing 4.5 (3.4) 4.0 (2.7) Aching 5.4 (2.4) 5.8 (2.1) Heavy 2.6 (3.2) 3.6 (2.7) Unpleasant 6.6 (1.8) 6.7 (1.6) Deep 6.7 (1.9) 6.8 (1.5) Surface 2.3 (2.8) 2.6 (2.0)
Improvement in Average Pain Intensity
Predicting Treatment Response ciated with a greater range of improvement from reports of worsening of activity/work interference by 6 points to improvement by 7.5 points), and higher baseline electric pain quality scores associated with only improvement ratings, ranging from 4 to 9 points for those whose baseline electric pain ratings were 7 or higher.
Patient Global Impression of Change
components retained a significant coefficient. We therefore considered primarily the significant model, which showed subjects with higher unpleasantness quality at baseline experiencing greater improvement in sleep interference (more reduced interference score) irrespective of treatment. This finding is illustrated in Figure 2. These data indicate that most participants found their pain to be unpleasant, with baseline scores for all but three subjects ranging from 5 to 10 on a 0 to 10 scale. Moreover, although most of the subjects reported decrease in sleep interference with treatment, those whose baseline unpleasantness score of 7 or greater evidenced a larger range in improvement (redetections ranging from 0 to 10 points), whereas those whose baseline unpleasantness ratings were 5 and 6 showed a more restricted range in improvement (reductions ranging from 0 to 7.5).
Improvement in Activity/Work Interference A similar pattern is seen when predicting interference with work and activities using electric and surface intensity pain qualities. The model is significant at the first step (R2 5 0.16, P < 0.05) with electric pain quality displaying a marginally significant coefficient weight (Beta 5 0.32, P 5 0.06) but surface intensity not being a significant predictor (P 5 0.47). The second step model showed similar results; the model was significant overall (R2 5 0.18, P < 0.05) with a nearly significant coefficient for electric quality (Beta 5 0.34, P 5 0.05) but nonsignificant coefficients for surface intensity or treatment condition (Beta 5 0.12 and 20.12, P 5 0.49 and 0.41, respectively). The model ceases to be significant when the interaction terms are added (overall R2 5 0.18, P 5 0.14) and no variable coefficients are significant when all of the variables are entered. The relationship of higher baseline electric pain quality scores to greater improvement in activity/work interference (more reduced interference score) is illustrated in Figure 3. As can be seen, low baseline electric pain quality scores are asso-
Discussion The study findings provide additional support for the ability of baseline pain quality measures to predict pain reduction following treatment, but extend the findings to include additional outcome measures, and indicate that different pain qualities may be associated with different outcomes. These results contribute to the growing body of research supporting the importance of examining pain qualities (in addition to pain intensity) in clinical and
Figure 3 Association between reduction in activity/work interference and baseline electric pain rating. 1337
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Figure 2 Association between reduction in sleep interference and baseline unpleasantness pain rating.
The model predicting PGIC scores from sensitive pain quality were similar to the pattern of results seen in predicting sleep and activity/work interference. PGIC assessed at post-treatment was significantly associated with the baseline sensitivity item (R2 5 0.11, P < 0.05), but subsequent steps adding treatment and the Treatment X Sensitivity ratings interaction resulted in a nonsignificant final model (P 5 0.14). At the first step the sole predictor, sensitive pain quality, had a coefficient beta of 0.32 (P < 0.05) showing subjects with higher scores in sensitive pain at baseline being more likely to report greater overall impressions of change (illustrated shown in see Figure 4). As can be seen, the pattern for findings for baseline sensitive pain predicting subsequent global ratings are much like the pattern found for electric pain predicting a decrease in activity/work interference; those with lower baseline sensitive scores evidence a larger range of outcomes, while those with higher baseline sensitive scores show both a reduced range in outcome, and an overall better response.
Gammaitoni et al.
research settings, and suggest those pain qualities that might be most important—at least in individuals with shoulder impingement pain. Our hypothesis that the pain qualities predicting treatment response would be similar across the two treatments examined (due to their similar mechanisms) was largely supported; in both treatment conditions, higher baseline levels of unpleasant pain, electric pain, and sensitive pain predicted a priori improvements in sleep interference, work/activity interference, and patient global ratings of improvement, respectively. In general, across these baseline measures, it was the participants who reported severe levels of these pain qualities (7 or higher on a 0–10 scale) that evidenced the most consistent improvement in outcome. These findings are consistent with the combined ideas that 1) those who have the most to gain (i.e., those reporting the highest baseline levels of various pain qualities) can expect the best response to effective treatments and 2) different pain qualities may be associated with different types of outcomes; specifically that pain unpleasantness may be more closely linked to sleep interference, electric pain sensations may be more closely linked to activity/work interference, and that sensitive pain may be most closely linked to global evaluations of the pain experience. In a previous study, we found that pain unpleasantness was among the strongest consistent predictors of sleep interference in three different samples of individuals with chronic pain [16]. The current findings are consistent with our earlier findings, and suggest that when pain is rated as severely unpleasant, even when or if it is not rated as severely intense, clinicians might pay particular attention to pain-related sleep problems as well. Given the importance of sleep to global health in general [17] as well as to pain management in particular [18], these patients may benefit from consideration of treatments for sleep problems, as appropriate. 1338
One pain quality was not a consistent predictor of treatment outcome across both interventions—hot pain. Moreover, the direction of associations found for this pain quality as a predictor of outcome differed from those found for the other predictors. Specifically, less hot pain at baseline for the participants who received the injection was (strongly) associated with more pain intensity reduction among the participants who received the injection only. Baseline levels of hot pain were only weakly (and not significantly) associated with treatment response to the patch. Again, given the fact that this study is the first time these associations have been examined for these treatments, it remains possible that the finding will not replicate. But the finding does suggest the possibility that despite similar hypothesized mechanisms of these treatment (i.e., blockade of sodium channels), the delivery method (injection versus patch) may influence outcome. Those with high hot pain, for example, might not benefit from the injection, whereas those with low hot pain could potentially benefit more from the injection. At this point, we are not recommending that treatment decisions be made based on the current findings alone; we are suggesting, however, that the findings strongly support the need for more research in this area to help determine which of the current findings replicate. The study has some important limitations that should be considered when interpreting the results. First, as already discussed, research examining the importance of baseline pain quality as a predictor of pain treatment outcome is still in its infancy. More studies evaluating a variety of treatments in a variety of pain populations are needed to help sort out which findings are most reliable. Only when multiple studies have been completed, will we be able to make empirically guided recommendations regarding the use of pain quality measures for guiding treatment. Related to this issue, there were a number of pain qualities that were relatively low at baseline. These included, for example, sharp, shooting, hot, and electric pain. Low baseline levels reduce the potential of variance due to floor effects, which can attenuate true associations that exist in the population. It is interesting, of course that two of these qualities, hot and electric, still emerged as significant predictors of
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
Figure 4 Association between patient global ratings of improvement and baseline sensitive pain rating.
To our knowledge, this study is the first that has examined the ability of baseline pain measures to predict treatment-related reductions in work and activity interference and patient global ratings of improvement. Thus, replication is needed before we can conclude that electric pain sensations and sensitive pain are specifically important to these outcomes. However, if replicated in samples from additional chronic pain populations, the findings would indicate particular importance for these pain qualities and domains, and perhaps a particular focus on identifying effective treatments for electric and sensitive pain, at least in order for treatment to also result in meaningful reductions of the extent to which pain interferes with function and the patient’s overall evaluation of treatment success.
Predicting Treatment Response
Despite the study limitations, however, the results provide further support for the potential of baseline pain qualities to predict who will response to different pain treatments (and how much). As such, they may ultimately provide an important clinical utility for helping guide treatment decisions if findings are replicated. The findings provide support for investigating this potential further.
Acknowledgments This study was supported by Nuvo. All authors are responsible for the work described in this paper, and they contributed to the conception, data interpretation, drafting of the manuscript, and revising the manuscript for intellectual content. All authors provided final approval of the version to be published.
References 1 Preston SJ, Arnold MH, Beller EM, Brooks PM, Buchanan WW. Variability in response to nonsteroidal anti-inflammatory analgesics: Evidence from controlled clinical therapeutic trial of flurbiprofen in rheumatoid arthritis. Br J Clin Pharmacol 1988;26:759–64. 2 Turk DC, Wilson HD, Cahana A. Treatment of chronic non-cancer pain. Lancet 2011;377:2226– 35. 3 Jensen MP, Galer BS, Gammaitoni AR, Lin CP. The Pain Quality Assessment Scale (PQAS) and Revised Pain Quality Assessment Scale (PQAS-R): Manual and user Guide, 2010. Available at: [email protected]. Last accessed March 18, 2015. 4 Jensen MP, Gammaitoni AR, Olaleye DO, Oleka N, Nalamachu SR, Galer BS. The pain quality assess-
ment scale: assessment of pain quality in carpal tunnel syndrome. J Pain 2006;7:823–32. 5 Gammaitoni AR, Smugar SS, Jensen MP, et al. Predicting response to pregabalin from pretreatment pain quality: Clinical applications of the pain quality assessment scale. Pain Med 2012;14:526– 32. 6 Radnovich R, Trudeau J, Gammaitoni A. A randomized clinical study of the heated lidocaine/ tetracaine patch versus subacromial corticosterioid injection for the treatment of pain assocaited with shoulder impingement syndrome. J Pain Res 2014; 7:727–35. 7 Watson CPN. Topical local anesthestics for neuropathic pain. In: Hansson PT, Fields HL, Hill RG, Marchettini P, editors. Neuropathic pain: Pathophysiology and treatment. Seattle: IASP Press; 2001: 215–21. 8 Turk DC, Dworkin RH, Allen RR, et al. Core outcome domains for chronic pain clinical trials: IMMPACT recommendations. Pain 2003;106:337– 45. 9 Jensen MP, Trudeau JJ, Radnovich R, Galer BS, Gammaitoni AR. The pain quality response profile of corticosteriod injections and heated lidocaine/tetracaine patch in the treatment of shoulder impingement syndrome. Clin J Pain, in press. 10 Victor TW, Jensen M, Gammaitoni AR, Gould EM, White RE, Galer BS. The dimensions of pain quality: Factor analysis of the Pain Quality Assessment Scale. Clin J Pain 2008;24:550–5. 11 Jensen MP, Karoly P. Self-report scales and procedures for assessing pain in adults. In: Turk DC, Melzack R, editors. Handbook of pain assessmenta New York: Guilford Press; 2011:19–44. 12 Turk DC, Dworkin RH, McDermott MP, et al. Analyzing multiple endpoints in clinical trials of pain treatments: IMMPACT recommendations. Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials. Pain 2008;139:485–93. 13 Dworkin RH, Turk DC, Farrar JT, et al. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. Pain 2005;113:9– 19. 14 Hosomi K, Shimokawa T, Ikoma K, et al. Daily repetitive transcranial magnetic stimulation of primary motor cortex for neuropathic pain: A randomized, multicenter, double-blind, crossover, shamcontrolled trial. Pain 2013;154:1065–72.
1339
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
treatment outcome despite their low baseline levels. This provides some support for the potential reliability of the findings related to these two pain qualities. It is also possible, even likely, that different pain qualities would predict outcome if different outcome measures were used. This idea is supported in part by our finding that different pain qualities predicted improvement in the four different outcomes measures examined here. Additional research would be needed to determine how pain assessed at baseline predict response not just to different treatments but as measured by different outcome variables. Finally, as noted previously, this analyses involved a large number of statistical tests without control for alpha inflation. Because of this, the findings should be considered exploratory and not confirmatory. Thus, additional research using larger sample sizes if possible and in also in samples of patients with more variability in baseline pain quality ratings would help to determine the reliability of these findings and may help identify additional predictors or allow for even stronger associations than those identified here.
Gammaitoni et al. 15 Rowbotham M, Harden N, Stacey B, Bernstein P, Magnus-Miller L. Gabapentin for the treatment of postherpetic neuralgia: A randomized controlled trial. JAMA 1998;280:1837–42. 16 Jensen MP, Dworkin RH, Gammaitoni AR, Olaleye DO, Oleka N, Galer BS. Do pain qualities and spatial characteristics make independent contributions to
interference with physical and emotional functioning? J Pain 2006;7:644–53. 17 Wells ME, Vaughn BV. Poor sleep challenging the health of a Nation. Neurodiagn J 2012;52:233–49. 18 Okifuji A, Hare BD. Do sleep disorders contribute to pain sensitivity? Curr Rheumatol Rep 2011;13:528–34.
Downloaded from http://painmedicine.oxfordjournals.org/ by guest on April 10, 2016
1340
