Abstract

Objective. The objective of this study was to compare the efficacy of gabapentin with placebo for neuropathic pain at the individual and population levels.

Design. This study used an n-of-1 trial methodology with three double-blind, randomized, crossover comparisons of gabapentin with placebo.

Setting. This study was carried out at specialist outpatient clinics at two Australian hospitals.

Patients. The patients are adults with chronic neuropathic pain.

Interventions. Following a dose-finding period, participants underwent three comparisons of 2-week periods on gabapentin (600–1,800 mg per day) and placebo. The dose-finding period was commenced by 112 patients, of whom 39 had no response so they did not enroll, leaving 73 trial participants. Of these, 48 completed and 7 partially completed their trials, and 18 withdrew.

Outcome Measures. The five outcome measures were the visual analog scale (0–10) of pain, sleep interference and functional limitation; frequency of adverse events and medication preference. The aggregate response was determined by weighting the response to each measure equally.

Results. Of the 55 participants who completed at least one cycle, the aggregate response to gabapentin was better than placebo in 16 (29%), of whom 15 continued gabapentin posttrial. No difference was shown in 38 (69%), and 1 (2%) showed a better response to placebo. Fifteen of these 39 continued gabapentin posttrial. Meta-analysis of the mean scores showed lower mean (standard deviation) scores for gabapentin by 0.8 (0.2) for pain, 0.6 (0.2) for sleep interference, and 0.6 (0.2) for functional limitation.

Conclusions. The response rate and mean reduction in symptoms with gabapentin were small. Gabapentin prescribing posttrial was significantly influenced by the trial results.

Introduction

Gabapentin is one of the many options for the pharmacological treatment of neuropathic pain, a condition characterized by neuronal hyperexcitability in damaged areas of the nervous system. There is level one evidence for the effectiveness of gabapentin in diabetic neuropathy, postherpetic neuralgia, and other neuropathic pain conditions [1,2]. The number needed to treat for diabetic neuropathy is 2.9 (confidence interval [CI] 2.2–4.3), for postherpetic neuralgia is 3.9 (CI 3.0–5.7), and for chronic neuropathic pain of all types is 4.3 (CI 3.5–5.7) [2]. While the results of randomized controlled trials (RCTs) and meta-analyses are useful in defining and predicting the mean clinical response to gabapentin at a population level, they comprise a wide range of responses by individuals, each of which is difficult to predict. Single patient or n-of-1 trials help to better define individual responses to treatment. N-of-1 trials are within-patient, randomized, double-blind, crossover comparisons, comparing a drug with a placebo (or another drug). They provide the most rigorous information available for any individual patient [3,4], and hence the potential to better target medication to responders. The results of a series of n-of-1 trials can also be meta-analyzed to give an estimate of response at a population level [5]. In the treatment of neuropathic pain, although there is a case report of an n-of-1 trial of a spinal cord stimulator [6], we have been unable to find any publications describing the use of n-of-1 trial methodology for the use of gabapentin.

We conducted a series of n-of-1 trials comparing gabapentin with placebo in patients with chronic neuropathic pain in the hospital setting to test the efficacy of gabapentin in individuals, and subsequently, at a population level by meta-analysis of completed trials. We also documented the changes in prescribing of pain medications that occurred with n-of-1 trials in the expectation that they would foster more appropriate, i.e., selective, use.

Methods

Trial Design

These n-of-1 trials were randomized, double-blind, crossover comparisons of gabapentin and placebo within individual participants. N-of-1 trials were offered to two groups of patients who had shown a clinical response to gabapentin—those who had completed a 2- to 3-week dose-finding period on gabapentin and those who were already on a maintenance dose of gabapentin. Each n-of-1 trial lasted 12 weeks, and consisted of three cycles of gabapentin and placebo treatment pairs assigned in random order, with each treatment lasting 2 weeks. Only measurements taken in the second week were used to allow for washout as the effects of gabapentin wear off within 2 days, as it has an elimination half-life of 5–7 hours (Figure 1). The choice of initial therapy was balanced in blocks of three to ensure that equivalent numbers started the study on either gabapentin or placebo. The participants, their doctors, and the dispensing pharmacies were all blinded to the identity and order of all treatment periods.

Figure 1

Flowchart of participants in n-of-1 trials of gabapentin vs placebo.

Figure 1

Flowchart of participants in n-of-1 trials of gabapentin vs placebo.

Participants and Ethics Approval

Participants came from specialist outpatient pain clinics in two Australian public hospitals, one in Port Kembla and the other in Brisbane. A few patients commenced as inpatients with follow-up as outpatients. Ethics approval was obtained from the relevant respective hospitals and university ethics committees.

Participants had to be willing to leave unchanged their usual medications, and their physical treatment and/or use of physical aids (all of which must have been used for the preceding 3 months). Informed consent was obtained from all participants.

Eligibility Criteria

Any adult patient with a clinical diagnosis of chronic neuropathic pain was eligible for inclusion in the study. Neuropathic pain was defined as pain initiated or caused by a primary lesion or dysfunction in the nervous system [7]. Specific conditions included phantom limb pain, poststroke pain, postherpetic neuralgia, diabetic neuropathy, and complex regional pain syndrome. Pain had to be of at least 3 months duration and of sufficient severity to warrant consideration of long-term gabapentin use. Patients were not admitted to the study if any of the following criteria were present: previous sensitivity to gabapentin; patients with a history of seizure; pregnancy; creatinine clearance <30 mL/min; certain H2 antagonists and pregabalin therapy; and episodic pain, for example, trigeminal neuralgia.

Medication

The open-label dose-finding period was 2–3 weeks. To follow usual clinical practice, patients were titrated to a maximum of 1,800 mg/day (600 mg three times daily), depending on response and adverse effects. This maximum dose was determined by the previous prescribing patterns for gabapentin at the respective hospitals. If there was no reduction in symptoms during the dose-finding period, the patient was not offered a trial. Patients who were already on gabapentin did not require a dose-finding period and were continued on their current doses. In the gabapentin periods of the n-of-1 trial, the gabapentin was commenced at 300 mg twice daily on day 1, and increased by 300 mg per day until the dose established in the dose-finding period was reached, either 600, 900, 1,200, or 1,800 mg per day.

The placebo capsules, manufactured by Arrow Pharmaceuticals (Croydon, Victoria, Australia), were identical to the corresponding active medication. Twelve-week medication kits of gabapentin and placebo capsules for each participant were prepared by a community pharmacy in 4-week lots, with 2 weeks of gabapentin and 2 weeks of placebo, the order allocated by a computer-generated randomization schedule by the University of Queensland n-of-1 trials service. Each 12-week kit was numbered and sent to the hospital pharmacies for dispensing in order of application by enrolling participants. After the n-of-1 trials, the participants were provided with gabapentin for 2 weeks while awaiting their results.

Many participants were also on stable doses of paracetamol, tricyclic antidepressants, nonsteroidal anti-inflammatory drugs, or opioids. Breakthrough/escape medication most commonly involved the addition of oral opioids. If debilitating pain was not controlled during a particular 2-week period, this period could be truncated and the next 2-week treatment period commenced.

Data Collection and Analysis

Descriptive information collected on each participant at entry included age, sex, duration and severity of chronic pain, degree of sleep interference, previous therapy, type of neuropathic pain, areas affected, and concomitant pain therapy. Each participant completed a diary with the following five outcome measures:

  • A daily estimation of pain in the marker area on a visual analog scale marked 0–10 (adapted from Daut et al. [8] and Bellamy et al. [9]) as the primary outcome measure.

  • A daily estimation of sleep interference on a visual analog scale marked 0–10.

  • A fortnightly Patient-Specific Function Scale adapted from Stratford et al. [10]. Participants rated, on a 0–10 visual analog scale, the degree to which up to five of their activities of daily living were limited by pain.

  • A fortnightly checklist of symptoms that could be adverse events, along with severity ratings.

  • A fortnightly global assessment of their medication preference, comparing their medication from the preceding fortnight with that from the fortnight before this.

Participants also recorded daily their use of analgesics or therapies other than the trial medications. Unused pill counts were used to monitored compliance.

After 12 weeks, all questionnaires were sent to the University of Queensland n-of-1 trial center for analysis. We averaged daily pain and sleep interference scores over the second week of each treatment period, discarding data from the first week to avoid carry-over effects. We calculated mean (standard deviation [SD]) differences in pain, sleep interference, and functional limitation scores between treatments for each of the three cycles, and from these calculated the adjusted mean difference for the three cycles using hierarchical Bayesian random-effects models. Like Zucker et al. [5], we assumed that the mean differences are normally distributed, and used their standard non-informative prior and hyperprior distributions. Calculations using Markov Chain Monte Carlo simulations were undertaken in WinBUGS version 1.4.1 (MRC Biostatistics Unit, University of Cambridge, UK) (http://www.mrc-bsu.cam.ac.uk/bugs). Simulations of size N = 10,000 were run after a burn-in period of 5,000 iterations, and convergence in the final samples was checked using visual plots of simulation histories.

The Bayesian method employed allows for individual assessment and group meta-analysis. The group meta-analysis could include data from all completed cycles, including from participants who completed less than three cycles. Assuming a minimum detectable difference of 1.0 for pain scores [11] and sleep interference scores [12], we defined a definite response as an adjusted mean absolute difference ≥1.0 less than the comparator, a probable response as a difference of ≥0.5 but <1.0 less than the comparator, and all other responses as no difference. Assuming a minimum detectable difference in functional limitation scores of 2.0 [13], we defined a definite response as an adjusted mean absolute difference ≥2.0 less than the comparator, a probable response as a difference of ≥1.0 but <2.0, and all other responses as no difference. As in Yelland et al. [14], we defined fewer adverse events or a preference for one medication as a definite response if they occurred in all three cycles, a probable response if they occurred in two cycles, and no response if they occurred in zero or one cycle.

Our summary outcome measure was an aggregate response score composed from an equally weighted combination of the five measures, each scored from −2 for a definite response to placebo to +2 for a definite response to gabapentin. This gave an aggregate score between −10 and 10. The absolute value of this score defined the response status with ≥6 considered a “definite response,”≥3 but <6 a “probable response,” and <3 a “non-response.”

A report was then sent to the referring doctor. After looking at the symptoms recorded, the doctor and participant decided together whether gabapentin was of greater benefit than placebo. After the n-of-1 trials, participants were interviewed by telephone about subsequent management decisions.

Results

Recruitment and Retention

Recruitment commenced in April 2004 and concluded in January 2006 due to reporting deadlines and funding constraints. One hundred twelve patients commenced the dose-finding period; 39 did not enroll in the trial due to a perceived lack of response to gabapentin (17), adverse events (12), or other reasons (10) (Figure 1). Seventy-three participants commenced n-of-1 trials. Fifty-five participants (75%) completed at least one cycle, allowing a partial analysis. Forty-eight participants (65%) completed three cycles. Two of these were participants who repeated their trials after disputing the negative results of their first trial. In both second trials, they showed a definite response to gabapentin, and this response was used as their definitive result.

Twenty-five participants (35%) withdrew during the trial period; reasons included excessive pain (8), side effects (6), unknown (6), inconvenience of trial (2), pain during placebo periods (2), and upcoming surgery (1).

Treatment Effects

The demographic and clinical characteristics of participants commencing trials appear in Table 1. Mean difference (SD) scores for pain, sleep interference, and functional limitation between gabapentin and placebo for each cycle are given in Table 2. The mean differences are all in the direction of lower scores with gabapentin, and are all 1 or less. Among the 55 participants who completed at least one cycle, the aggregate response of the five outcomes showed a definite response to gabapentin in 8 (15%), and a probable response in 8 (15%), giving a total of 16 positive responders (29%) (Table 3). There was no response to gabapentin in 38 (69%), and a probable response to placebo in 1 (2%). The response rate for participants on 1,800 mg per day was significantly higher than for those taking less than 1,800 mg per day (48% vs 19%, χ2 = 5.3, degree of freedom [df] = 1, P = 0.02). Response rates for individual outcomes of pain, sleep interference, and adverse events resembled the aggregate response rates closely but differed somewhat for functional limitation where there were no definite responders and for medication preference where there were 11 definite responders.

Table 1

Demographic and clinical characteristics of completers/partial-completers and total commencers of the gabapentin n-of-1 trials

Variable Completers/Partial Completers (N = 55) Non-Completers (N = 18) Total (N = 73) 
Age in years—mean (minimum, maximum) 57.8 (24, 94) 56.9 (31, 88) 57.6 (24, 94) 
Duration of pain (years)    
  ≤1     4 (7%)     5 (28%)     9 (12%) 
  2–5   28 (51%)     6 (33%)   34 (47%) 
  6–10     8 (15%)     1 (6%)     9 (12%) 
  >10   11 (20%)     3 (17%)   14 (19%) 
  Unknown     4 (7%)     3 (17%)     7 (10%) 
Gender    
  Male   18 (33%)   12 (66%)   30 (41%) 
  Female   37 (67%)     6 (34%)   43 (59%) 
Pretrial main regular medication    
  Gabapentin     9 (16%)     2 (11%)   11 (15%) 
  Antidepressant     4 (7%)     0 (0%)     4 (5%) 
  NSAIDs     4 (7%)     2 (11%)     6 (8%) 
  Opioids   10 (18%)     4 (22%)   14 (5%) 
  Anticonvulsant     5 (9%)     1 (6%)     6 (8%) 
  Paracetamol     6 (11%)     5 (28%)   11 (15%) 
  Paracetamol plus codeine     8 (15%)     1 (6%)     9 (12%) 
  Other/unknown     2 (4%)     3 (17%)     5 (7%) 
  No drug     7 (13%)     0 (0%)     7 (10%) 
Daily gabapentin dose during trial (mg)    
  600     6 (10%)     1 (6%)     7 (9%) 
  900   13 (24%)     7 (39%)   20 (27%) 
  1,200   12 (22%)     4 (22%)   16 (22%) 
  1,800   23 (42%)     6 (33%)   29 (40%) 
  Unknown     1 (2%)     0 (0%)     1 (1%) 
Variable Completers/Partial Completers (N = 55) Non-Completers (N = 18) Total (N = 73) 
Age in years—mean (minimum, maximum) 57.8 (24, 94) 56.9 (31, 88) 57.6 (24, 94) 
Duration of pain (years)    
  ≤1     4 (7%)     5 (28%)     9 (12%) 
  2–5   28 (51%)     6 (33%)   34 (47%) 
  6–10     8 (15%)     1 (6%)     9 (12%) 
  >10   11 (20%)     3 (17%)   14 (19%) 
  Unknown     4 (7%)     3 (17%)     7 (10%) 
Gender    
  Male   18 (33%)   12 (66%)   30 (41%) 
  Female   37 (67%)     6 (34%)   43 (59%) 
Pretrial main regular medication    
  Gabapentin     9 (16%)     2 (11%)   11 (15%) 
  Antidepressant     4 (7%)     0 (0%)     4 (5%) 
  NSAIDs     4 (7%)     2 (11%)     6 (8%) 
  Opioids   10 (18%)     4 (22%)   14 (5%) 
  Anticonvulsant     5 (9%)     1 (6%)     6 (8%) 
  Paracetamol     6 (11%)     5 (28%)   11 (15%) 
  Paracetamol plus codeine     8 (15%)     1 (6%)     9 (12%) 
  Other/unknown     2 (4%)     3 (17%)     5 (7%) 
  No drug     7 (13%)     0 (0%)     7 (10%) 
Daily gabapentin dose during trial (mg)    
  600     6 (10%)     1 (6%)     7 (9%) 
  900   13 (24%)     7 (39%)   20 (27%) 
  1,200   12 (22%)     4 (22%)   16 (22%) 
  1,800   23 (42%)     6 (33%)   29 (40%) 
  Unknown     1 (2%)     0 (0%)     1 (1%) 
Table 2

Mean difference and standard deviation (SD) for pain, sleep interference, and functional limitation scores between gabapentin and placebo over each cycle (N = 55)

Scores Cycle 1
 
Cycle 2
 
Cycle 3
 
Mean Difference SD Mean Difference SD Mean Difference SD 
Pain 0.6 1.8 0.60 2.3 0.8 2.4 
Sleep interference 0.5 1.7 0.53 2.1 0.4 2.4 
Functional limitation 0.6 2.6 0.41 2.1 1.0 2.4 
Scores Cycle 1
 
Cycle 2
 
Cycle 3
 
Mean Difference SD Mean Difference SD Mean Difference SD 
Pain 0.6 1.8 0.60 2.3 0.8 2.4 
Sleep interference 0.5 1.7 0.53 2.1 0.4 2.4 
Functional limitation 0.6 2.6 0.41 2.1 1.0 2.4 

A positive mean difference signifies a lower (more favorable) score for gabapentin than for placebo, whereas a negative mean difference signifies a lower (more favorable) score for placebo than for gabapentin.

Table 3

Response status in participants completing at least one cycle for each of the five outcomes and the aggregate response weighting all available outcomes equally (N = 55)

 Gabapentin
 
No Difference Placebo
 
Definitely Better Probably Better Probably Better Definitely Better 
Pain scores 10 10 33 
Sleep interference scores   6   8 40 
Functional limitation scores   1   7 47 
Preferred medication 11   5 37 
Adverse events   6   7 36 
Aggregate response   8   8 38 
 Gabapentin
 
No Difference Placebo
 
Definitely Better Probably Better Probably Better Definitely Better 
Pain scores 10 10 33 
Sleep interference scores   6   8 40 
Functional limitation scores   1   7 47 
Preferred medication 11   5 37 
Adverse events   6   7 36 
Aggregate response   8   8 38 

Hierarchical Bayesian meta-analysis of the mean scores of participants completing at least one cycle showed lower mean-adjusted (SD) scores for gabapentin by 0.8 (0.2) for pain, 0.6 (0.2) for sleep interference, and 0.6 (0.2) for functional limitation.

There were many changes in pain medications from before to after the trial. Of the 55 participants completing trials, 27 (49%) were prescribed gabapentin after their trial. Of the 15 taking gabapentin before their trial, 8 (53%) continued gabapentin afterward. Of the 40 not taking gabapentin before their trial, 23 (58%) were prescribed gabapentin afterward.

The prescribing decisions about gabapentin immediately after the trial were consistent with the trial result in 33 (60%), inconsistent in 17 (31%), and unknown in five (9%). Gabapentin was continued in 15 (94%) of the 16 responders in comparison with 15 (38%) of the 39 non-responders (χ2 = 14.0, df = 1, P < 0.001).

Discussion

In the present study, we used an n-of-1 trial to identify responders and target prescribing. In this cohort, only 29% of participants showed a positive response to gabapentin. Although the n-of-1 trial design does not allow a calculation of the “number needed to treat,” it is possible to calculate the “number needed to trial” to find one positive responder as four. This number is surprisingly high given that the use of a dose-finding period for new patients and the inclusion of patients already taking gabapentin may have increased the response rate by excluding many patients with adverse events or no subjective response. In contrast, the ceiling dose of 1,800 mg, compared with 3,600 mg in other studies, may have reduced response rates, as there is evidence from this and another study of reduced effectiveness in trials using lower doses [1].

The ability to generalize the results of this series of trials to the broader population of patients with neuropathic pain is limited by participant selection methods, the limited ceiling dose, and the withdrawal rate. The population included patients with a clinical diagnosis of chronic neuropathic pain from a variety of causes.

For pain and sleep interference scores, the observed mean differences for each cycle were similar to the mean-adjusted differences in the meta-analysis of all cycles but were considerably less than those reported in other RCTs of gabapentin [15,16]. This is likely due to issues of external validity or generalizability, with n-of-1 patients more likely to arise from a more diverse population pool (e.g., more likely to have multiple comorbidities) than those eligible for the RCTs.

The withdrawal rate of 35% was high, but this is fairly typical of n-of-1 trials [3,17,18] and indeed typical of many conventional RCTs [19]. The shift in response status from negative to positive in the two participants who repeated trials suggests either problems with the reproducibility of the method or the possibility of unblinding by the participants determined to show a positive result to support continuation of their gabapentin.

Neuropathic pain is among the most severe and difficult to manage of all chronic pains. Management involves a multifaceted approach, including the use of various medicines, physiotherapy, and behavioral modification. These were left unchanged in the present study, with only gabapentin or placebo as the introduced variables. This approach improves the generalizability of our results to clinical practice compared with neuropathic pain RCTs in which the experimental agent is used as monotherapy.

Although gabapentin has become established as a treatment for neuropathic pain, it is a relatively expensive medication. It cannot be generally subsidized in Australia for neuropathic pain without data from comparative head-to-head studies to show better value than cheaper alternates. In the Australian health system, it is only subsdized for pain in hospital-based outpatient clinics and not in general practice, resulting in much of the cost burden for gabapentin falling on the hospital sector. To target long-term prescribing for a chronic condition such as neuropathic pain, this n-of-1 study provided high-level evidence to inform decisions about prescribing in the hospital sector. Almost every responder to gabapentin was continued on this medication following their trial. However, 38% of non-responders were continued on gabapentin, suggesting that there were other factors influencing decisions to prescribe gabapentin. Some of these may have been the different weightings doctors and participants placed on each outcome compared with the aggregate response that was based on an equal weighting of five outcomes. Further qualitative investigation is required to understand the factors influencing prescribing for chronic neuropathic pain.

Acknowledgments

Many thanks to Donna-Marie Preston and Grace McBride from the University of Queensland, and Tracey Jordan from Port Kembla Hospital for collecting the data, and the doctors and patients who participated. Thank you to the Princess Alexandra Hospital, Brisbane, and the Port Kembla Hospital, Port Kembla, for providing staff, infrastructure, and permission for the study to be conducted in their clinics.

This study was supported by a grant from the Australian Health Ministers Advisory Council. The Australian Health Ministers Advisory Council had no role in the study design, data collection, analysis and interpretation, or writing of the article. Nor did they control or influence the decision to submit the final manuscript for publication. The placebo capsules were provided by the Illawara Area Health Service. The researchers are independent from the funders. No conflict of interest is declared.

Contributors

Michael Yelland took overall responsibility for the study design and conduct, and contributed to the data analysis and preparation of the article. He is also the study's guarantor.

Chris Poulos, Peter Pillans, Guy Bashford, and Joanna Sturtevant contributed to the study design and conduct, and contributed to the data analysis and preparation of the article.

Jane Nikles and Chris Del Mar obtained the original grant, contributed to the study design and conduct, and contributed to the data analysis and preparation of the article.

Norma Vine coordinated the study, contributed to data collection and data analysis, and had input into the article.

Phillip Schluter provided statistical advice, set up the Bayesian analysis, and had input into writing the article.

Robyn Brown contributed to the study design, and data collection and analysis.

Meng Tan, Jonathan Chan, and Fraser Mackenzie contributed to the study conduct and data collection analysis.

References

1
Mellegers
MA
Furlan
AD
Mailis
A
.
Gabapentin for neuropathic pain: Systematic review of controlled and uncontrolled literature
.
Clin J Pain
 
2001
;
17
(
4
):
284
95
.
2
Wiffen
PJ
McQuay
HJ
Edwards
JE
Moore
RA
.
Gabapentin for acute and chronic pain
.
Cochrane Database Syst Rev
 
2005
;(
3
):
cD005452
. DOI: .
3
Guyatt
GH
Keller
JL
Jaeschke
R
et al
The n-of-1 randomized controlled trial: Clinical usefulness. Our three-year experience
.
Ann Intern Med
 
1990
;
112
:
293
9
.
4
Nikles
CJ
Clavarino
AM
Del Mar
CB
.
Using n-of-1 trials as a clinical tool to improve prescribing
.
Br J Gen Pract
 
2005
;
55
:
175
80
.
5
Zucker
DR
Schmid
CH
McIntosh
MW
et al
Combining single patient [N-of-1] trials to estimate population treatment effects and to evaluate individual patient responses to treatment
.
J Clin Epidemiol
 
1997
;
50
:
401
10
.
6
Cepeda
MS
Acevedo
JC
Miranda
N
Cortes
C
Carr
DB
.
An N-of-1 trial as an aid to decision-making prior to implanting a permanent spinal cord stimulator
.
Pain Med
 
2008
;
9
(
2
):
235
9
.
7
Merskey
H
Bogduk
N
.
Classification of Chronic Pain: Descriptions of Chronic Pain Syndromes and Definitions of Pain Terms
 .
Seattle, WA
:
IASP Press
;
1994
:
212
.
8
Daut
RL
Cleeland
CS
Flanery
RC
.
Development of the Wisconsin Brief Pain Questionnaire to assess pain in cancer and other disease
.
Pain
 
1983
;
17
:
197
210
.
9
Bellamy
N
Buchanan
WW
Goldsmith
CH
Campbell
J
Stitt
LW
.
Validation study of WOMAC: A health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee
.
J Rheumatol
 
1988
;
15
:
1833
40
.
10
Stratford
PW
Gill
C
Westaway
M
Binkley
J
.
Assessing disability and change in individual patients: A report of a patient-specific measure
.
Physiother Can
 
1995
;
47
:
258
63
.
11
Ehrich
EW
Davies
GM
Watson
DJ
et al
Minimal perceptible clinical improvement with the Western Ontario and McMaster Universities osteoarthritis index questionnaire and global assessments in patients with osteoarthritis
.
J Rheumatol
 
2000
;
27
(
11
):
2635
41
.
12
Zisapel
N
Nir
T
.
Determination of the minimal clinically significant difference on a patient visual analog sleep quality scale
.
J Sleep Res
 
2003
;
12
(
4
):
291
8
.
13
Chatman
AB
Hyams
SP
Neel
JM
et al
The Patient-Specific Functional Scale: Measurement properties in patients with knee dysfunction
.
Phys Ther
 
1997
;
77
:
820
9
.
14
Yelland
MJ
Nikles
CJ
McNairn
N
et al
Celecoxib compared with sustained-release paracetamol for osteoarthritis: A series of n-of-1 trials
.
Rheumatology (Oxford)
 
2007
;
46
(
1
):
135
40
.
15
Backonja
M
Beydoun
A
Edwards
KR
et al
Gabapentin for the symptomatic treatment of painful neuropathy in patients with diabetes mellitus: A randomized controlled trial
.
JAMA
 
1998
;
280
(
21
):
1831
6
.
16
Rice
AS
Maton
S
.
Gabapentin in postherpetic neuralgia: A randomized, double blind, placebo controlled study
.
Pain
 
2001
;
94
(
2
):
215
24
.
17
Larson
EB
.
N-of-1 clinical trials. A technique for improving medical therapeutics
.
West J Med
 
1990
;
152
:
52
6
.
18
Pincus
T
Koch
G
Lei
H
et al
Patient preference for placebo, acetaminophen or celecoxib efficacy studies [PACES]: Two randomized placebo-controlled cross-over clinical trials in patients with osteoarthritis of the knee or hip
.
Ann Rheum Dis
 
2004
;
63
:
931
9
.
19
Martin
R
Vogtle
L
Gilliam
F
Faught
E
.
Health-related quality of life in senior adults with epilepsy: What we know from randomized clinical trials and suggestions for future research
.
Epilepsy Behav
 
2003
;
4
:
626
34
.