The utilization of data retrieval strategies in clinical trials
Mouna Akacha, Novartis Pharma AG  Chengxing Lu, Novartis Pharmaceutical Corporation  *David Ohlssen , Novartis Pharmaceuticals Corporation 

Keywords: Retrieved data, Treatment switching, Target estimand, Sensitivity analysis

There has been a long debate in the statistical literature on missing data issues and health authorities are increasingly concerned about impact of missing data on statistical inference for clinical trials. In 2010, the Committee for Medicinal Products for Human Use (CHMP) published a guidance document on missing data in confirmatory clinical studies. At the same time, the National Research Council published a report on ‘The Prevention and Treatment of Missing Data in Clinical Studies’ which was commissioned by the U.S. Food and Drug Administration. Both documents recommend the collection and use of retrieved data, i.e. data collected after study-treatment discontinuation.

Retrieved data are frequently used in primary analyses for outcome trials. However, this is less the case for symptomatic studies where the drug effect is expected to be lost quite quickly. This is because it is common practice for dropout patients to switch therapies after study treatment discontinuation. Treatment differences may thus be inappropriately exaggerated or minimized and the underlying question of interest may be modified.

In this talk, we will discuss statistical and organizational aspects which need to be considered when using retrieved data. We will also discuss alternative approaches which enable investigation of ‘off-treatment after study-treatment discontinuation’ scenarios.