References 1. Hadzic A, Vloka J. Peripheral Nerve Blocks: Principles and Prac- tice. New York, NY: McGraw-Hill; 2004. 2. Symes A, Ellis H. Variations in the surface anatomy of the spinal accessory nerve in the posterior triangle. Surg Radiol Anat 2005;27:404-408. 3. Winnie AP. Interscalene brachial plexus block. Anesth Analg 1970;49:455-466. 4. Bodner G, Harpf C, Gardetto A, Kovacs P, Gruber H, Peer S, Mallhoui A. Ultrasonography of the accessory nerve: Nor- mal and pathologic findings in cadavers and patients with iatrogenic accessory nerve palsy. J Ultrasound Med 2002; 21:1159-1163. 5. Lu L, Haman SP, Ebraheim NA. Vulnerability of the spinal accessory nerve in the posterior triangle of the neck: A cadaveric study. Orthopedics 2002;25:71-74. Accepted for publication December 14, 2006. doi:10.1016/j.rapm.2006.12.003 Intra-Articular Morphine in Acute Pain Trials To the Editor: We thank the editor for the chance to respond to the stimulating editorial by Professor Cristoph Stein, who com- mented on some issues concerning our article on intra- articular (IA) morphine. 1,2 In our study, we found no an- algesic effect of 5 mg of IA morphine compared with placebo when given to patients with moderate or severe postoperative pain. Professor Stein disagrees with our inter- pretation. We agree with Professor Stein that significant baseline pain is necessary when the efficacy of a single analgesic drug dose is compared with placebo in randomized clin- ical trials (RCT). A single-dose comparative analgesic RCT should include only patients with at least moderate pain, preferably moderate-to-severe pain. 3,4 Detection of sta- tistically significant group differences in pain intensity (or pain relief) is the primary outcome measure. A significant problem with most published IA morphine RCTs is that the patients are included at the end of surgery, before any pain can be experienced or measured. This study design invariably will include many patients without pain. The fraction of patients who experience significant pain after knee arthroscopy was unknown until we documented that the incidence of at least moderate pain was only 57% in male and 84% in female patients, 5 a gender difference that is statistically and clinically significant. The fact that the patients after knee arthroscopy fall into 1 of these 2 groups (1 group who never experience more than mild pain and 1 group who experience moderate pain or more) represents a serious methodological problem in studies in which the test medication is given before baseline pain can be assessed. Inclusion and randomization of all patients at the end of surgery may lead to a false-positive pain outcome if the sample size is small (most IA morphine RCTs have group size of 20 or less). 6 Trials that result in subsequent high mean pain intensity in the placebo group may indicate sensitivity, but they may also indicate that more patients with pain, by random variation, were ran- domized to the placebo group and fewer patients with pain were randomized to the treatment group. This action could have induced bias in IA morphine trials. Thus, pain intensity above 30 mm VAS in the placebo group does not prove assay sensitivity. This outcome is not a problem when only pa- tients with at least moderate pain before the test interven- tion are included in RCTs. In our last IA morphine RCT, we included only patients with moderate-to-severe pain before test drugs were ad- ministered. 1 This trial would have been one of the largest IA morphine trials with a pre-emptive design if we had included all 60 patients in the trial and injected the test drugs at the end of the arthroscopic procedure. Sample size is important, but it is not the only way to increase validity. Inclusion of patients with moderate-to-severe pain (and exclusion of patients with no or mild pain) has been documented to increase assay sensitivity. 7 Internal sensitivity has been proved in the study design used in our recent publication. 8 Excluding patients with no, or only mild, pain increases mean pain intensity, and vari- ance (SD) is greatly reduced. Patients with symptoms or clinical signs of acute arthri- tis (rheumatoid arthritis or acute inflammation after trauma) were not included in our trial. In the population screened for this trial, we did not encounter any patients with these conditions. The patients are representative of day-case arthroscopic patients in Norway and are proba- bly not very different from day-case knee patients in the western part of the world. We had 1 patient with acute arthritis in a previous intra-articular morphine trial and excluded him from a subgroup analysis of synovial-fluid inflammatory mediators, but the patient was included in the analysis of analgesic efficacy and side effects. 9 Sub- group analysis of inflammatory mediators documents in- flammation in all patients, and we found significantly higher levels of prostaglandin E 2 in patients with moder- ate pain compared with patients who experienced no pain or mild pain. 10 Thus, the population in the study is not in any way special, even in the degree of inflamma- tion. The failure to reject the null hypothesis in this RCT is of course not evidence for equivalence between mor- phine and saline, but if any effect of IA morphine occurs, the effect size must be small and not clinically significant for acute pain after arthroscopic procedures. A possible role for peripheral opioids in chronic pain states should be studied further and is not contradicted by our study. Professor Stein suggests that “to find the ‘truth,’ the reader will have to go back to the original literature and delve into details such as inclusion and exclusion criteria, raw pain scores, absolute dosages, type and combination of supplemental analgesics, and statistical analysis of the data.” We fully agree. 6,11 Leiv Arne Rosseland, M.D. Nina Solheim, M.D. Department of Anesthesiology Rikshospitalet Radiumhospitalet Medical Center Oslo, Norway 176 Regional Anesthesia and Pain Medicine Vol. 32 No. 2 March–April 2007