Estimation of clinical trial success rates and related parameters

Introduction

• The probability of success (POS) of a clinical trial is critical for clinical researchers and biopharma investors to evaluate when making scientific and economic decisions.

• When the POS is not estimated regularly there is a possibility of misjudgment of the risk and value of the drug development. This causes opportunities to be lost for both investors and patients.

• the big challenge in estimating a clinical trails success rate is the access to accurate information on trial characteristics and outcomes

  • this process is expensive, time consuming, and susceptible to error
  • trained analysts would require many hours of labour to manually produce POS estimates

paper shows estimates of the POS and risk related characteristics of clinical trails using 406 038 entries of industry- and non-industry-sponsored trials, corresponding to 185 994 unique trials over 21 143 compounds from Informa Pharma Intelligence’s Trialtrove and Pharmaprojects databases from January 1, 2000 to October 31, 2015.

• to process big amounts of data, an automated algorithm traces the path of drug development, infers phase transitions, and computes POS stats (this is done in hours)

. Apart from the gains in efficiency, our algorithmic approach allows us to perform previously infeasible computations, such as generating time-series estimates of POS and related parameters.

• What is estimated: aggregated success rates, completion rates, phase transition probabilities, trial durations, and other disaggregated measures (clinical phases, disease, types of organization, if biomarkers are used)

Data

• data: citeline data provided by Informa Pharma Intelligence

  • data: multiple data streams (example ClinicalTrials.gov, press releases, financial reports, study reports, drug marketing, label applications, secondary sources from consulting companies)
  • note that secondary sources are often used to reduce biases that may come from the tendency of organizations to report only successful trials

• a single trial can be repeated as multiple data points

• Some trials are missing end-dates due to the failure of companies to report this info