Identifying Predictors of Response Using an Individual Patient Data Meta-Analysis

Abstract

Osteoarthritis is a common painful condition affecting the joints. Currently, there is no cure for osteoarthritis, only the management of its symptoms. To personalize pain relief to an individual, we need to first identify predictors of treatment response. We undertook two individual patient data (IPD) meta-analyses conducted as part of the OA Trial Bank. One aim was to identify predictors of response to two topical treatments: topical non-steroidal anti-inflammatory drugs and capsaicin. The other aim was to identify placebo responders and predictors of response in osteoarthritis. IPD meta-analysis requires the original trial authors to share the data on all the individual patients in their trial. The raw data are then re-analyzed centrally while preserving the clustering of participants within their original trials. Obtaining the IPD from randomized controlled trials is the most important step in conducting an IPD meta-analysis. However, collecting IPD is a long process and a flexible approach to data collection is needed. The advantages of an IPD meta-analysis includes an increased study sample size, while allowing both individual patient-level and study-level predictors of response to be taken into account. Meanwhile, being restricted only to variables previously measured within the trials is a limitation. In this case study, we report on some of the challenges we experienced, and lessons learnt from undertaking two different IPD meta-analyses to identify predictors of response in patients with osteoarthritis.

locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles