Quality assessment A commonly used three-item, five-point quality scale was used to rate the quality of the trials [Jadad et al,1996]. The minimum score possible for inclusion of a study in the review was 2 (one point each for randomisation and double blinding). The maximum score possible was 5 (2 points for descriptions of randomisation, 2 points for descriptions of double blinding, and 1 point for descriptions of withdrawals). Jadad AR, Moore RA, Carroll D et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17:1-12. How points are awarded: 1) Is the study randomised? If yes, + 1 point. Is the randomisation procedure appropriate and reported in the study? If yes, +1 point. If no, delete all points awarded for randomisation. 2) Is the study double blind? If yes, + 1 point. Is the double blinding method appropriate and reported in the study? If yes, +1 point. If no, delete all points awarded for double blinding. 3) Are the reasons for patient withdrawals and dropouts described, for each treatment group? If yes, +1 point.Validity assessment The Oxford Pain Validity Scale was used to rate the validity of trials [Smith et al, 2000]. This is a 16-point scale; details are shown in the Table. Smith LA, Oldman AD, McQuay HJ, Moore RA. Teasing apart quality and validity in systematic reviews: an example from acupuncture trials in chronic neck and back pain. Pain 2000; 86:119-32.The Oxford Pain Validity Scale (OPVS) should be only be used on trials which are: * RANDOMISED * have a start group size ≥10 for all groups relevant to the review question ITEM SCORE (circle one number per item) COMMENTS 1. Blinding Was the trial convincingly double-blind? Was the trial convincingly single-blind or unconvincingly double-blind? Was the trial either not blind or the blinding is unclear? 6 3 0 i.e. states double-blind and how this was achieved, eg doubledummy, identical appearance, etc. i.e. states single-blind and how this was achieved, eg observerblind, patient-blind, etc. 2. Size of trial groups Was the start group start size ≥40? Was the start group start size 30 to 39? Was the start group start size 20 to 29? Was the start group start size 10 to 19? 3 2 1 0 Not all groups in the trial will necessarily be relevant to the review question. Rate this item using the smallest group that is relevant to the review question. 3. Outcomes Look at pre hoc list of most desirable outcomes relevant to the review question: Did the paper include results for at least one pre hoc desirable outcome, and use the outcome appropriately? There were no results for any of the pre hoc desirable outcomes, or, a pre hoc desirable outcome was used inappropriately. 2 0 NB: If the trial has not reported the results of any measures relevant to the review question (even if it described them in methods) it should be excluded from the review. 4. Demonstration of internal sensitivity Look at the baseline levels for the outcomes relevant to the review question: For all treatment groups, baseline levels were sufficient for the trialist to be able to measure a change following the intervention (eg enough baseline pain to detect a difference between baseline and post-treatment levels). Altertatively, did the trial demonstrate internal sensitivity? For all treatment groups, baseline levels were insufficient to be able to measure a change following the intervention, or, baseline levels could not be assessed, or internal sensitivity was not demonstrated. 1 0 One way to demonstrate internal sensitivity is by having an additional active control group in the trial which demonstrates a significant difference from placebo (ie the trial design is able to detect a difference). For example, by having an extra group treated with an anlagesic known to be statistically different from placebo, and by demonstrating this difference. Alternatively, internal sensitivity can be demonstrated with a dose response. i) Definition of outcomes Did the paper define the relevant outcomes clearly, including where relevant, exactly what 'improved', 'successful treatment', etc represented? The paper failed to define the outcomes clearly. 1 0 There must be at least one outcome measure defined clearly to score 1. This item refers to any outcome measure relevant to the review question, not just pre hoc desirables. ii) Data presentation: Location and dispersion Did the paper present either mean data with standard deviations, or dichotomous outcomes, or median with range, or sufficient data to enable extraction of any of these? The paper presented none of the above. 1 0 iii) Statistical Testing Did the trialist chose an appropriate statistical test, with correction for multiple tests where relevant? Inappropriate statistical tests were chosen and/or multiple testing was carried out, but with no correction, or, no statistics were carried out. 1 0 Corrections for multiple testing must be put in place when a series of tests or measures have been carried out on the same patient group. iv) Handling of Dropouts The dropout rate was either ≤10%, or was >10% and includes an intention-to-treat analysis in which dropouts were included appropriately. The dropout rate was >10% and dropouts were not included in the analysis, or, it is not possible to calculate a dropout rate from data presented in the paper. 1 0 TOTAL SCORE Author conclusion: Trial positive / Trial negative Reviewer conclusion: Trial positive / Trial negative 5. Data Analysis The Oxford Pain Validity Scale (OPVS)