POS1361 THE INFLUENCE OF BASELINE DEMOGRAPHICS ON VARIABILITY OF OA PAIN ASSESSED BY WOMAC PAIN CHANGE FROM BASELINE IN INTERVENTIONAL TRIALS

Background Development of new products for OA treatment is difficult, partly due to excessive placebo response and high variability in patient-reported pain outcomes, and exclusion of trial subjects with potentially confounding pain from non-target joints may increase the effect size. This may in part be due to reductions in variability, and hence in standard deviation (SD) of the mean changes in pain scores, which improves the statistical power to detect differences between study groups. The impact of refining the study population in terms of added benefit to the SD and statistical power vs. added screen failures is poorly described. Objectives To investigate the impact of common demographic and pain characteristics on the SD of mean pain change from baseline in a large interventional OA trial database. Methods Data from 2 randomized controlled trials of oral salmon calcitonin in OA (1) were analyzed post-hoc. The SD of the mean change from baseline (SD-CFB) to Year 2 in the WOMAC knee pain score (0-100) was calculated. The SD-CFB was also calculated for subgroups (e.g. demographics, pain of the target and non-target knee). The sample size required to identify a statistically significant difference of at least 8 out of 100 between groups (with 80% power) was calculated, and the additional proportion of eligible subjects to be screen-failed if the particular subgroup was excluded was calculated to quantify the potential impact on the study feasibility. Results A total of 1,487 subjects had WOMAC pain data throughout the trial period. Results are shown in Table 1. Few clinical characteristics influenced the SD and hence the required sample size, except for Asian race, associated with a lower SD compared to Caucasians (19.71 vs. 21.85), to detect a significant difference. Exclusion of those with high BMI (≥ 35 kg/m2) led to a reduced sample size required of 113. Subjects with non-target knee pain above the median had higher pain variability and thus a larger sample sizes required to detect differences vs. those with pain scores below the median. However, exclusion of these groups required screen failing approx. extra 50 % of the eligible population to achieve this. A less exclusive method was excluding subjects with non-target knee pain not exceeding that of the target knee. Radiographic characteristics did not influence the SD, except for those with a KL grade 3 or 4 of the non-target knee (Table 1). This observation requires further scrutiny. Conclusion High BMI and high baseline pain of the non-target knee may contribute negatively to the study power, however, the added study cost in terms of additional screened subjects required to replace those excluded based on these parameters should be carefully evaluated. The potential impact of these parameters on the magnitude of the difference between study groups was not evaluated and may also differ, with additional statistical implications. Reference [1]Karsdal MA, Byrjalsen I, Alexandersen P, Bihlet A, Andersen JR, Riis BJ, Bay-Jensen AC, Christiansen C. Treatment of symptomatic knee osteoarthritis with oral salmon calcitonin: results from two phase 3 trials. CSMC021C2301/2 investigators. Osteoarthritis Cartilage 2015;23:532-543. Acknowledgements: NIL. Disclosure of Interests Asger Reinstrup Bihlet Employee of: Full-time employee and shareholder of NBCD, Peter Alexandersen Employee of: Part-time employee at NBCD, Jeppe Ragnar Andersen Employee of: Full-time employee and shareholder of NBCD, Yanqi Li Employee of: Full-time employee and shareholder of NBCD, Morten Karsdal Employee of: Full-time employee and shareholder of NBCD, Claire Prener Miller Employee of: Full-time employee at NBCD.