IMPORTANT

This document has been supplied under Pfizers copyright licences. It must not be used for promotional purposes.

This document is protected by copyright and it may not (even for internal purposes) be further copied, stored or on-copied electronically without permission, save as may be permitted by law.

The recipient may print out a single paper copy of any document received electronically.

ORIGINAL PAPER

Evaluation of the bromyalgia impact questionnaire at baseline as a predictor for time to pain improvement in two clinical trials of pregabalin
A. G. Bushmakin,1 J. C. Cappelleri,1 A. B. Chandran,2 G. Zlateva2

Pzer Inc., Statistics, Groton, CT, USA Pzer Inc., Outcomes Research, New York, NY, USA Correspondence to: Joseph C. Cappelleri, Pzer Inc., 445 Eastern Point Road (MS 8260-2502), Groton, CT 06340, USA Tel.: +1 860 441 8033 Fax: +1 860 686 5139 Email: joseph.c.cappelleri@ pzer.com Disclosures Andrew G. Bushmakin, Joseph C. Cappelleri, Arthi B. Chandran, and Gergana Zlateva are employees and stockholders of Pzer Inc, the sponsor of this study.

SUMMARY Background: The Fibromyalgia Impact Questionnaire (FIQ) is a patient-reported outcome that evaluates the impact of bromyalgia (FM) on daily life. This study evaluated the relationships between the functional status of FM patients, measured with the FIQ at baseline, and median time to a clinically relevant pain reduction. Methods: Data were derived from two randomised, placebo-controlled trials that evaluated pregabalin 300, 450 and 600 mg day for the treatment of FM. The KaplanMeier (nonparametric) method was applied to estimate median times to transient and stable events. The transient event was dened as a 27.9% improvement on an 11-point daily pain diary scale (0 = no pain, 10 = worst possible pain), and the stable event was dened as the mean of the daily improvements 27.9% relative to baseline over the subsequent study duration starting on the day of the transient event. A parametric model using time-to-event analysis was developed for evaluating the relationship between baseline FIQ score and the median time to these events. Results: Median time was longer among patients treated with placebo relative to pregabalin for the transient events (1112 days vs. 57 days) and stable events (86 days vs. 1329 days). A signicant association was observed between baseline FIQ scores and median time to transient and stable events (p < 0.001). Median times to events were similar between the studies. For transient pain reduction events, median times ranged from 3.0 to 4.5 days for baseline FIQ scores of 10, and 9.19.6 days for FIQ scores of 100; for stable pain reduction events, the median time ranged from 11.0 to 13.0 days and from 27.0 to 28.5 days for baseline FIQ scores of 10 and 100 respectively. Conclusions: Time to a clinically relevant reduction in pain was signicantly associated with FM severity at baseline as measured by the FIQ. Such an analysis can inform patient and physician expectations in clinical practice.

Whats known
Categorisation of bromyalgia as a multidimensional condition has enabled development of patient-reported outcome measures that have successfully evaluated the disease and its treatment within the context of pain (with a numeric rating scale, NRS) and, separately, patient function and disease severity surrounding the impact of bromyalgia on daily life (with the Fibromyalgia Impact Questionnaire, FIQ).

Whats new
Among subjects with bromyalgia, longer median times to pain reduction (as measured by NRS), both transient and stable reduction, after pregabalin treatment were associated with higher levels of baseline bromyalgia severity (as measured by the FIQ). Determining the time to clinical efcacy and its relationship with baseline disease severity is a useful approach for enhancing management strategies by informing patients and physicians regarding clinically realistic expectations of pregabalin treatment.

Introduction
Fibromyalgia (FM) is a chronic condition that may occur in individuals regardless of age or gender, but disproportionately affects women in an approximate ratio of 7 : 1 (1). Although FM has been conventionally characterised by chronic widespread pain, fatigue and multiple tender points (2), the frequent presence of a constellation of comorbid symptoms contribute to the reduced function and quality of life reported by patients with FM (35). These symptoms include sleep disturbances, headaches, irritable bowel, cognitive dysfunction, anxiety and depression, all of which are associated with signicant impact on daily functioning. It is now recognised that these symptoms

are part of the clinical presentation of FM (6) and have been incorporated into clinical guidance (7). The aetiology and pathogenesis of FM have yet to be fully elucidated. Nevertheless, re-evaluation of the clinical concept of FM, incorporating results of neuroimaging studies, support a dysfunction in central pain processing and other alterations in neurotransmitters as integral to the pathways underlying the multidimensional nature of this disease (8,9). Recent evidence-based recommendations for the management of FM suggest the use of non-pharmacologic and pharmacologic therapies (10), with the latter consisting of agents from diverse therapeutic classes. Although these classes include analgesics, antidepressants, anticonvulsants, muscle relaxants

52

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259. doi: 10.1111/ijcp.12035

Time to meaningful pain reduction

53

and sedatives hypnotics, only three drugs have been specically approved for the treatment of FM: pregabalin (11), duloxetine (12) and milnacipran (13). In contrast with the antidepressants duloxetine and milnacipran, pregabalin is an anticonvulsant that is a high afnity ligand of alpha2-delta subunits of voltage-gated calcium channels (14). Pregabalin has been shown to reduce the release of neurotransmitter, including glutamate (15,16) and substance P (17). Pregabalin was the rst pharmacologic therapy approved in the US for the treatment of FM, and has also been approved at doses up to 450 mg daily for FM in 24 countries, including Canada, Russia, Israel, and several Asian, Middle Eastern and Latin American countries. Its approval was based on demonstration of statistically and clinically signicant reductions in pain relative to placebo as well as improvements in other patient-reported outcomes (PROs) (1820). Although pain remains a primary outcome measure in FM, the categorisation of FM, as a multidimensional condition, has enabled development of PROs that evaluate FM within a broader context of patient functioning and categorisation of disease severity (21). In particular, the Fibromyalgia Impact Questionnaire (FIQ) was developed and validated as a PRO for evaluating the impact of FM on daily life (22). Cutoff scores for dening severity levels on the FIQ have been established, as has a clinically relevant difference on the FIQ (23). In clinical trials, PROs, such as the FIQ, are generally analysed in the form of the differences between treatment arms. However, time-to-event modelling (i.e. survival analysis) can provide an additional perspective of treatment effects that includes initial improvement and stable improvement. Such a perspective may enhance our knowledge of the relevant events that may be expected during treatment in a clinical setting, including time to improvement and an understanding of why patients switch medications, especially if a surrogate marker is identied that can be used to estimate the time to response. The purpose of the current investigation is to evaluate the relationship between functional impairment in patients with FM, measured by FIQ scores at baseline, and the median time-to-improvement for clinically meaningful reductions in pain for different doses of pregabalin and placebo.

Methods
Data used for modelling in the current analysis were derived from two similarly designed randomised, placebo-controlled trials that evaluated three doses of pregabalin (300, 450 and 600 mg day) for the treat-

ment of FM (19,20). It should be noted that 600 mg day is not an approved or a recommended dose for FM. All patients, including patients randomised to 600 mg day, were included in this secondary analysis to be consistent with the two main efcacy publications emanating from the two studies (19,20). In one study (Study 1), 745 patients were randomised to 14 weeks of double-blind treatment (19), and in the other study (Study 2), 748 patients were randomised to 13 weeks of treatment (20). Both studies were conducted in accordance with the Declaration of Helsinki, approved by the appropriate institutional review boards or independent ethics committees, and patients provided written informed consent prior to participation. Details of the subjects, methodology and results of the trials have been previously published (19,20). Two types of events were considered in this analysis, a transient event and a stable event. The transient pain reduction event was dened as an improvement by 27.9% on an 11-point daily pain diary numerical rating scale (0 = no pain, 10 = worst possible pain); this percent reduction in pain has been previously determined to be a clinically important difference (24). The stable pain reduction event was dened as having the mean of the daily improvements be 27.9% relative to baseline over the subsequent study duration starting on the day of the transient event. Two different methods of time-to-event analysis were used to provide distinct sets of results (25). In the rst time-to-event analysis, KaplanMeier (nonparametric) methodology (26) was applied to estimate median time to events stratied by treatment arm. In the second analysis, parametric models were t to evaluate the relationship between baseline FIQ scores and the median time to event for patients treated with pregabalin. The models included time to event as the outcome accounting for censoring, and baseline FIQ as the predictor. Time to event was taken to follow a gamma distribution, and thus the gamma model provides the parametric survival estimate. The model included an intercept parameter, the coefcient parameter for baseline FIQ, and scale and shape parameters. On the basis of the model, the times to transient and stable events were predicted for deciles of baseline FIQ values. Scoring of the FIQ is on a 0100 scale, with higher scores indicating a greater impact of FM. Using the same types of model, a secondary analysis was performed that evaluated baseline pain as a predictor of time to event, with the times to transient and stable events predicted for each baseline pain severity score from 4 to 10; a minimum score

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259

54

Time to meaningful pain reduction

of 4 was required for entry into the clinical trials. All analyses were conducted using sas version 9.2 (SAS Institute Inc., Cary, NC, USA).

Results
Subject demographics in both studies were similar, well-balanced across treatment groups, and consistent with what may be expected for an FM population (Table 1). The overall population was predominantly female (94.4%) and white (90.6%), with a mean age of 49.4 (standard deviation 11.2) years. On the basis of KaplanMeier estimation, median times to the transient event of clinically meaningful
Table 1 Demographic and clinical characteristics of the study populations

pain reduction were similar in both studies (Table 2), and were longer among patients treated with placebo (1112 days) relative to pregabalin (57 days). Median times to stable pain reduction events were also similar between the studies; pregabalin resulted in a shorter median time relative to placebo, with a median time for placebo subjects in study 1 that was beyond the study duration. Although the pregabalin 300 mg dose generally resulted in a longer median time to event, there was a general lack of signicance in the median time to transient and stable events across active treatment dose groups, enabling the doses to be pooled in each study for the predictive modelling using a parametric approach. In addition, the proportions of patients

Placebo Study 1 (n = 184) Study 2 (n = 190)

Pregabalin 300 mg Study 1 (n = 183) Study 2 (n = 185)

Pregabalin 450 mg Study 1 (n = 190) Study 2 (n = 183)

Pregabalin 600 mg Study 1 (n = 188) Study 2 (n = 190)

Variable

Sex, n (%) Female Male Age, years, mean (SD) Race, n (%) White Black Other FM duration, years, mean (SD) Baseline pain score, mean (SD) Baseline FIQ score, mean (SD)

169 (91.8) 15 (8.2) 49 (11.4) 169 7 8 10.3 6.6 58.7 (91.8) (3.8) (4.3) (9.0) (1.3) (15.6)

183 (96.3) 7 (3.7) 48.6 (11.3) 167 10 13 8.8 7.2 65.1 (87.9) (5.3) (6.8) (6.9) (1.2) (13.7)*

173 (94.5) 10 (5.5) 49.1 (11.2) 164 9 10 9.6 6.7 61.1 (89.6) (4.9) (5.5) (7.0) (1.3) (15.7)

174 (94.1) 11 (5.9) 50.1 (10.4) 169 10 6 9.6 7.1 65.8 (91.4) (5.4) (3.2) (8.6) (1.4) (13.4)*

183 (96.3) 7 (3.7) 50.8 (11.8) 171 12 7 10.3 6.6 59.6 (90.0) (6.3) (3.7) (7.7) (1.4) (15.1)

169 (92.3) 14 (7.7) 47.7 (10.8) 169 7 7 9.6 7.1 63.6 (92.3) (3.8) (3.8) (8.5) (1.4) (13.9)*

179 (95.2) 9 (4.8) 50.9 (11.1) 174 5 9 9.9 6.7 59.5 (92.6) (2.7) (4.8) (8.3) (1.4) (16.2)

180 (94.7) 10 (5.3) 48.7 (11.2) 170 8 12 9.3 7.0 62.7 (89.5) (4.2) (6.3) (7.6) (1.1) (13.2)*

*Calculated separately as they were not reported in the study publication.

Table 2 Median time to events based on KaplanMeier estimation

Median time to event, days (percent responders) Study Placebo Pregabalin 300 mg Pregabalin 450 mg Pregabalin 600 mg

p* All subjects Pregabalin subjects

Transient event Study 1 12 Study 2 11 Stable event Study 1 Study 2 86

(75.3) (75.5) (42.3) (46.8)

7 (83.5) 7 (79.4) 23 (61.5) 29 (59.4)

6 (85.7) 6 (87.8) 14 (63.0) 18 (62.8)

6 (86.5) 5 (84.7) 15 (58.9) 13 (60.3)

0.0001 < 0.0001 < 0.0001 0.0029

0.6818 0.0271 0.725 0.7005

*On the basis of the log-rank test of equality among distributions of the treatment groups examined; p-values for all subjects indicate that at least one intervention group was different from at least another intervention group, and p-values for pregabalin subjects indicates that at least one pregabalin group was different from at least one other pregabalin group for transient event in study 2. Transient event dened as a 27.9% improvement on an 11-point daily pain diary numerical rating scale (0 = no pain, 10 = worst possible pain). Stable event dened as the mean of the daily improvements being maintained at a level 27.9% relative to baseline over the study duration starting on the day of the transient event.

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259

Time to meaningful pain reduction

55

who were transient and stable responders were consistently higher with pregabalin (Table 2), and these proportions were similar between the two studies. For the pooled doses of pregabalin in the parametric model, a clear relationship was observed in both studies between baseline FIQ scores and median time to improvement for the transient event of pain reduction (Figure 1A, C) and the stable event of pain reduction (Figure 1B, D). These graphs show that longer median times to improvement were associated with higher (less favourable) FIQ scores at baseline, and these relationships were statistically signicant (p < 0.001). On the basis of this model, the median times to the transient event of pain reduction across deciles of baseline FIQ scores were generally consistent between Study 1 and Study 2 (Table 3), and increased from 4.5 [95% condence interval (CI) 3.4, 6.0] and 3.0 (95% CI 2.1, 4.3) days for baseline FIQ scores of 10 in Study 1 and 2, respectively, to 9.1 (95% CI 7.2, 11.5) and 9.6 (95% CI 7.5, 12.4), respectively, for baseline scores of 100. Similarly, the median times to stable events were also consistent between the studies and were approximately three times higher than the

median to transient events, regardless of the FIQ score. For the stable event, the estimated median time for subjects with baseline FIQ scores of 10 was 13.0 (95% CI 8.6, 19.7) and 11.0 (95% CI 6.6, 18.3) in Studies 1 and 2, respectively, and this increased to 28.5 (95% CI 19.5, 41.5) and 27.0 (95% CI 18.1, 40.4), respectively, for subjects with baseline FIQ scores of 100. Clear relationships were also observed between baseline NRS pain score and time to transient and stable events (Figure 2). Estimation of median time to events was consistent between the two studies for each level of pain severity (Table 4), and there was a moderate correlation between baseline pain and baseline FIQ; Pearson r = 0.52 and 0.51 in Study 1 and Study 2 respectively (p < 0.0001 for both studies).

Discussion
Although the results reported in this study were derived from clinical trials, they are relevant to the clinical setting by providing data that can inform physicians and patients regarding treatment expectations, such as the length of time to onset of efcacy.

Figure 1 Parametric model demonstrating the relationship between baseline score on the Fibromyalgia Impact

Questionnaire (FIQ) and predicted median time to events. Black circles represent predicted median time to event and the shading represents the 95% condence interval. Transient event in Study 1 (A) and Study 2 (C) dened as a 27.9% improvement on an 11-point daily pain diary numerical rating scale (0 = no pain, 10 = worst possible pain). Stable event in Study 1 (B) and Study 2 (D) dened as the mean of the daily improvements being maintained at a level 27.9% relative to baseline over the study duration starting on the day of the transient event.

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259

56

Time to meaningful pain reduction

Table 3 Predicted median and 95% condence intervals (95% CI) for time to event in active treatment arms in Studies 1 and 2 as a function of baseline scores of the Fibromyalgia Impact Questionnaire (FIQ).

Median (95% CI) time to event, days Baseline FIQ score Transient event* Study 1 Study 2 Stable event Study 1 Study 2

10 20 30 40 50 60 70 80 90 100

4.5 4.9 5.2 5.7 6.1 6.6 7.2 7.8 8.4 9.1

(3.4, (3.8, (4.3, (4.9, (5.4, (5.9, (6.3, (6.7, (6.9, (7.2,

6.0) 6.2) 6.4) 6.6) 7.0) 7.4) 8.1) 9.1) 10.2) 11.5)

3.0 3.4 3.9 4.4 5.1 5.8 6.5 7.4 8.5 9.6

(2.1, (2.6, (3.1, (3.7, (4.4, (5.2, (5.8, (6.4, (7.0, (7.5,

4.3) 4.6) 5.0) 5.3) 5.8) 6.4) 7.3) 8.6) 10.3) 12.4)

13.0 14.2 15.5 16.9 18.4 20.1 21.9 23.9 26.1 28.5

(8.6, 19.7) (9.9, 20.5) (11.2, 21.5) (12.6, 22.7) (14.1, 24.2) (15.4, 26.2) (16.7, 28.9) (17.7, 32.3) (18.7, 36.4) (19.5, 41.5)

11.0 12.2 13.5 14.9 16.4 18.1 20.0 22.2 24.5 27.0

(6.6, 18.3) (7.9, 18.9) (9.3, 19.5) (10.9, 20.4) (12.5, 21.5) (14.1, 23.3) (15.5, 25.9) (16.6, 29.6) (17.4, 34.3) (18.1, 40.4)

*Transient event dened as a 27.9% improvement on an 11-point daily pain diary numerical rating scale (0 = no pain, 10 = worst possible pain). Stable event dened as having the mean of the daily improvements be 27.9% relative to baseline over the subsequent study duration starting on the day of the transient event.

Figure 2 Parametric model demonstrating the relationship between baseline pain Numerical Rating Scale (NRS; 0 = no

pain, 10 = worst possible pain) score and predicted median time to events. Black circles represent predicted median time to event and the shading represents the 95% condence interval. Transient event in Study 1 (A) and Study 2 (C) dened as a 27.9% improvement on the NRS. Stable event in Study 1 (B) and Study 2 (D) dened as the mean of the daily improvements being maintained at a level 27.9% relative to baseline over the study duration starting on the day of the transient event.

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259

Time to meaningful pain reduction

57

Table 4 Predicted median and 95% condence intervals (95% CI) for time to event in active treatment arms in Studies 1 and 2 as a function of baseline pain scores on a Numerical Rating Scale (NRS).

Median (95% CI) time to event, days Baseline NRS pain score Transient event* Study 1 Study 2 Stable event Study 1 Study 2

4 5 6 7 8 9 10

4.6 5.3 6.0 6.9 7.9 9.0 10.4

(3.7, (4.5, (5.3, (6.2, (6.9, (7.5, (8.2,

5.6) 6.1) 6.8) 7.7) 9.1) 10.8) 13.1)

3.5 4.2 5.0 6.0 7.1 8.5 10.2

(2.8, (3.6, (4.4, (5.4, (6.3, (7.3, (8.4,

4.2) 4.9) 5.6) 6.6) 8.0) 10.0) 12.5)

14.8 16.6 18.6 20.7 23.2 25.9 29.0

(10.7, (12.4, (14.2, (15.9, (17.4, (18.8, (20.0,

20.7) 22.2) 24.3) 27.1) 30.9) 35.8) 42.1)

13.0 14.6 16.5 18.7 21.0 23.8 26.8

(9.2, 18.2) (10.9, 19.6) (12.7, 21.5) (14.5, 24.0) (16.2, 27.3) (17.7, 31.8) (19.1, 37.5)

*Transient event dened as a 27.9% improvement on an 11-point daily pain diary numerical rating scale (0 = no pain, 10 = worst possible pain). Stable event dened as having the mean of the daily improvements be 27.9% relative to baseline over the subsequent study duration starting on the day of the transient event.

These results show that there is a direct and signicant relationship between FM severity, assessed using the FIQ, and the time to a clinically signicant reduction in pain with pregabalin; a longer time to pain reduction was observed at greater levels of FM severity. Furthermore, the magnitude of both the relationship and the reduction can be quantied in a meaningful, practical and comprehensive manner. The results of these analyses, with respect to both responder rates and response times, were consistent across both studies, demonstrating stability and repeatability. In particular, the substantial rates of response with pregabalin suggest that the median time to event is unlikely to be inuenced by patients who failed to achieve the study endpoint. Importantly, both models (FIQ- and pain-based) resulted in predictions that were similar. Furthermore, the models based on different studies also gave close results, providing further support of the models stability. Patients who have not (yet) experienced the event are said to be censored. Censored data contribute valuable information and they should not be omitted from the analysis. It would also be wrong to treat the observed time (at censoring) as the survival time. In our study, we used a specialised set of statistical methods that have been developed for handling such data. These methods use all information up to the time of censoring and do not throw away information. The censored observations contribute to the total number at risk up to the time that they ceased to be followed. If more than 50% of the observations are censored, estimates may not be reliable (27). In our study, less than 50% of the subjects (about 40%) did not experience a stable event (and hence were

censored) and, therefore, we expect the estimates produced to be reliable. If the subjects were to experience a stable event, it would be after their censored time and the analysis makes use of this information. Approximately 75% of patients on placebo were responders in the transient time-to-event analysis. This high percentage is suggestive of a large placebo effect for a transient response and underscores the importance of also conducting an analysis where responders are to have a stable event, which invokes a more rigorous standard. In the stable time-to-event analysis, the percentage of subjects on placebo who responded falls substantively to 42% in one study and 47% in another. In supplementing the transient analysis, the stable event analysis therefore adds fair balance to the overall analysis. As indicated by the general lack of signicance across active treatment dose groups in the time-toevent curves for transient and stable events, the dose of pregabalin was consistently observed to be a statistically insignicant factor in the onset and maintenance of efcacy. Pregabalin 300 mg day resulted in a longer numerical estimate of median time to event relative to the other doses [450 mg day is the maximum recommended dose for the treatment of FM (28)], which suggests a possible doseresponse relationship. Across all doses of pregabalin, subjects with mild FM (FIQ score < 39) had a median time to initial efcacy of approximately 45 days, whereas those with moderate (FIQ score 39 to < 59) and severe FM (FIQ score 59) had median times of approximately 6 days and 79 days respectively. Pooling the three doses of pregabalin increased the stability of the results.

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259

58

Time to meaningful pain reduction

Although a short time to onset of efcacy after initiation of therapy is of great benet, the ability to maintain a stable level of improvement during treatment is an equally important attribute that contributes to patterns of dosing, adherence and switching. In the current analysis, the median time to stable improvement was approximately three times longer than the time to onset, regardless of baseline FM severity (as measured by FIQ) or pain severity (as measured by NRS Pain). Two previous studies of pregabalin estimated the time to onset of pain relief (29,30). In these studies, however, the type of population and the denition of pain reduction differed from the ones presented in the current study. Patients in the previous two studies had postherpetic neuralgia or painful diabetic neuropathy; moreover, pain relief was dened as a 1-point reduction in pain relative to baseline for those patients who had clinically meaningful pain reduction (> 30%) at end point (29,30). One of these studies reported a median time to pain reduction of 1.5 days for xed-dose pregabalin in patients with postherpetic neuralgia (29), and the other reported pain reductions by 2 days after initiating treatment (30). However, in contrast with those studies, which considered a statistically signicant reduction in pain, the current analysis used a clinically signicant reduction in pain as the marker for efcacy and evaluated the entire pregabalin cohort regardless of response. Therefore, both the median time to transient event and the median time to stable event may represent more realistic and relevant estimates of what may be expected in clinical practice. An important limitation of this study is that it was based on clinical trial populations, which may be different from that of the clinical setting in that they may reect patients who are more motivated or adherent to therapy. Thus, in clinical practice, it is possible that the time to achieve signicant and consistent improvements may take longer than reported herein, especially since time for titration of dose to therapeutic levels may vary among treating physicians. Another potential limitation is that the predictive model for baseline pain only covered the NRS range of 410, whereas for the FIQ, the full-scale range was used. However, a minimum pain score of 4 was part of the entry criteria for the clinical trials but there was no minimum FIQ score. It should also be noted that although pain is a primary characteristic of FM and remains a key objective of therapy, the FIQ provides a more comprehensive and clinically relevant perspective by assessing FM and categorising its severity as a condition. The moderate correlation between baseline FIQ and pain scores further supports the concept of FM as a condition that is characterised by factors in addition to pain.

This study could also potentially be criticised for not evaluating the time to withdrawal because of adverse events. Among the patients treated with pregabalin in both studies, adverse events and withdrawals because of adverse events, which occurred in up to 32.6% of patients, were dose-dependent, with dizziness and somnolence the adverse events that most frequently led to withdrawal (19,20). However, it should be considered that while a patient may or may not discontinue therapy upon occurrence of an adverse event, in the absence of knowledge of when efcacy is likely to occur, patients may discontinue prior to its onset. Thus, informing patient and physician expectations of the timing of treatment benets may be more clinically relevant for maintaining persistence with therapy than occurrence of adverse events. As such, this study focused on determining efcacy onset and whether time to clinically meaningful reductions in pain are associated with, and can be predicted by pain and functional impairment at baseline; similar evaluations for adverse events may be worth pursuing in a subsequent analysis. Despite these limitations, this study represents a new approach to evaluating clinical trial data that may have direct application to clinical practice.

Conclusions
This study demonstrated how techniques for time-toevent analysis can be used to explore relationships between baseline values of a surrogate measure and efcacy outcomes. In particular, for patients with FM, baseline FIQ scores were predictive of the time to efcacy in a clinically meaningful and quantiable manner. Such an understanding has practical applications by informing patients and physicians regarding the clinical expectations of therapy and enhancing management strategies. This relationship may also provide a metric for comparative analysis of expected outcomes and, in addition, a methodological framework for evaluating changes in other outcomes as a function of a baseline surrogate measure.

Acknowledgements
This study was funded by Pzer Inc. Editorial assistance was provided by E. Jay Bienen and was funded by Pzer, Inc.

Author contributions
All authors contributed to the study design, statistical analysis plan, results interpretation and review of the draft manuscript; the nal manuscript was read and approved by all authors.

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259

Time to meaningful pain reduction

59

References
1 Wolfe F, Ross K, Anderson J et al. The prevalence and characteristics of bromyalgia in the general population. Arthritis Rheum 1995; 38: 1928. 2 Wolfe F, Smythe HA, Yunus MB et al. The American College of Rheumatology 1990 Criteria for the Classication of Fibromyalgia. Report of the Multicenter Criteria Committee. Arthritis Rheum 1990; 33: 16072. 3 Arnold LM, Crofford LJ, Mease PJ et al. Patient perspectives on the impact of bromyalgia. Patient Educ Couns 2008; 73: 11420. 4 Hoffman DL, Dukes E. The health status burden of people with bromyalgia: a review of studies that assessed health status with the SF-36 or the SF-12. Int J Clin Pract 2008; 62: 11526. 5 Choy E, Perrot S, Leon T et al. A patient survey of the impact of bromyalgia and the journey to diagnosis. BMC Health Serv Res 2010; 10: 102. 6 Fitzcharles MA, Yunus MB. The clinical concept of bromyalgia as a changing paradigm in the past 20 years. Pain Res Treat 2012; 2012: 184835. Epub 2011 Oct 29. Published online 2011 October 29. doi: 10.1155/2012 184835. 7 Wolfe F, Clauw DJ, Fitzcharles M-A et al. The American College of Rheumatology preliminary diagnostic criteria for bromyalgia and measurement of symptom severity. Arthritis Care Res 2010; 62: 60010. 8 Clauw DJ, Arnold LM, McCarberg BH, FibroCollaborative. The science of bromyalgia. Mayo Clin Proc 2011; 9: 90711. 9 Ceko M, Bushnell MC, Gracely R. Neurobiology underlying bromyalgia symptoms. Pain Res Treat 2012; 2012: 585419. Epub 2011 Oct 27. Published online 2011 October 27. doi:10.1155/2012 585419. 10 Carville SF, Arendt-Nielsen S, Bliddal H et al. EULAR evidence based recommendations for the 11 12

13

14

15

16

17

18

19

20

management of bromyalgia syndrome. Ann Rheum Dis 2008; 67: 53641. Pzer Inc. Lyrica [Pregabalin] Capsules Prescribing Information. New York, NY: Pzer, Inc., 2011. Eli Lilly and Company. Cymbalta [Duloxetine Hydrochloride] Delayed Release Capsules Prescribing Information. Indianapolis, IN: Eli Lilly and Company, 2011. Forest Pharmaceutical. Savella [Milnacipran Hydrochloride] Prescribing Information. St. Louis, MO: Forest Pharmaceuticals, 2009. Field MJ, Cox PJ, Stott E et al. Identication of the alpha2-delta-1 subunit of voltage-dependent calcium channels as a molecular target for pain mediating the analgesic actions of pregabalin. Proc Natl Acad Sci USA 2006; 103: 1753742. Errante LD, Petroff OA. Acute effects of gabapentin and pregabalin on rat forebrain cellular GABA, glutamate, and glutamine concentrations. Seizure 2003; 12: 3006. Cunningham MO, Woodhall GL, Thompson SE et al. Dual effects of gabapentin and pregabalin on glutamate release at rat entorhinal synapses in vitro. Eur J Neurosci 2004; 20: 156676. Maneuf YP, Hughes J, McKnight AT. Gabapentin inhibits the substance P-facilitated K(+)-evoked release of [(3)H]glutamate from rat caudial trigeminal nucleus slices. Pain 2001; 93: 1916. Crofford LJ, Rowbotham MC, Mease PJ et al. Pregabalin for the treatment of bromyalgia syndrome: results of a randomized, double-blind, placebo-controlled trial. Arthritis Rheum 2005; 52: 126473. Arnold LM, Russell IJ, Diri EW et al. A 14-week, randomized, double-blind, placebo-controlled monotherapy trial of pregabalin in patients with bromyalgia. J Pain 2008; 9: 792805. Mease PJ, Russell IJ, Arnold LM et al. A randomized, double-blind, placebo-controlled, phase III trial of pregabalin in the treatment of patients with bromyalgia. J Rheumatol 2008; 35: 50214.

21 Boomershine CS. A comprehensive evaluation of standardized assessment tools in the diagnosis of bromyalgia and in the assessment of bromyalgia severity. Pain Res Treat 2011; 2012: 653714. Epub 2011 Oct 10. 22 Burckhardt CS, Clark SR, Bennett RM. The bromyalgia impact questionnaire: development and validation. J Rheumatol 1991; 18: 72833. 23 Bennett RM, Bushmakin AG, Cappelleri JC et al. Minimal clinically important difference in the bromyalgia impact questionnaire. J Rheumatol 2009; 36: 130411. 24 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: 14958. 25 Kleinbaum DG, Klein M. Survival Analysis. A SelfLearning Text. 2nd edn. New York, NY: Springer Science + Business Media, Inc., 2005. 26 Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 45781. 27 Lagakos SW. General right censoring and its impact on the analysis of survival data. Biometrics 1979; 35: 139156. 28 Pzer Inc. Lyrica [Pregabalin] Capsules Prescribing Information. New York, NY: Pzer, Inc., 2009. 29 Stacey BR, Barrett JA, Whalen E et al. Pregabalin for postherpetic neuralgia: placebo-controlled trial of xed and exible dosing regimens on allodynia and time to onset of pain relief. J Pain 2008; 9: 100617. 30 Sharma U, Griesing T, Emir B, Young JP Jr. Time to onset of neuropathic pain reduction: a retrospective analysis of data from nine controlled trials of pregabalin for painful diabetic peripheral neuropathy and postherpetic neuralgia. Am J Ther 2010; 17: 57785. Paper received April 2012, accepted August 2012

2012 Blackwell Publishing Ltd Int J Clin Pract, January 2013, 67, 1, 5259