Risk ratios (RRs) and 95% CIs were used to summarize contingency table

results for OMT vs. sham OMT for the following primary outcome measures: initial clinical response, stable clinical response (vs. never-response), relapse, and clinical response at the week 12 exit visit. Subgroup analyses based on patient characteristics and use of Crizotinib purchase non-prescription and prescription medication for LBP during the trial were conducted for each primary outcome. A P-value for interaction was computed to assess the statistical significance of differences between subgroups ( Altman and Bland, 2003). The Cochrane Back Review Group criteria were used to interpret the magnitude of treatment effects (i.e., effect sizes) for OMT based on the relevant RR. For statistically significant results pertaining to clinical response, treatment effects were classified buy Ion Channel Ligand Library as large (RR > 2), medium (1.25 ≤ RR ≤ 2), or small (RR < 1.25). Correspondingly, for relapse, treatment effects were classified as large (RR < 0.5), medium (0.5 ≤ RR ≤ 0.8), or small (RR > 0.8) (Furlan et al., 2009). Analyses were primarily performed using intention-to-treat methods with per-protocol analyses available for further assessment of study findings. Hypotheses were tested using a 0.05 level of statistical significance with

the SPSS Statistics version 21 software package (IBM Corporation, Armonk, NY). The flow of patients through the trial is illustrated in the CONSORT diagram (Fig. 1). A total of 186 patients with high baseline pain severity were randomized, including 95 patients assigned to OMT and 91 patients assigned to sham OMT. Overall, the median age of patients was 43 years (IQR, 22 years) and 115 (62%) patients were women. The median baseline VAS pain score was 63 mm (IQR, 16 mm). A total of 103 (55%) patients reported LBP for more than one year, although relatively few patients had ever been hospitalized

or had surgery for LBP. Co-morbid depression was reported by 46 (25%) patients. There was no significant difference between treatment groups in any baseline patient characteristic (Table 1). Patients in the OMT group were more likely to be treated by faculty physicians than fellows or residents (P = 0.04) and more frequently developed Interleukin-3 receptor a contraindication to continued trial participation (P = 0.03) than patients in the sham OMT group. There was no significant difference between treatment groups in any other measure of patient follow-up or treatment adherence. Clinical response and relapse profiles over time for each patient revealed important differences between the OMT and sham OMT groups (Fig. 2). The weighted proportion of time wherein patients experienced substantial LBP improvement was 0.39 (95% CI, 0.32–0.47) for patients who received OMT vs. 0.20 (95% CI, 0.14–0.26) for patients who received sham OMT (P < 0.001).