Predictive approach to 'Quality' in clinical trials
Authors: Deepak P.N., AVP - Principal Technology Architect; Pooja Durgad, Senior Associate Consultant; Renuka Natarajan, Senior Associate Consultant
Quality is not an act. It is a habit ~ Aristotle
A recent engagement with the clinical trials quality team of one of the top 10 pharmaceutical companies, located in North America, gave more insights to the way sponsors handle and manage the risks throughout the lifecycle of a typical clinical trial. Quality Risk management or QRM is the process of proactively identifying risks, prioritizing the risks identified and coming up with a mitigation plan to reduce the errors that matter for the success of the clinical trial. QRM is a collaborative effort between different stakeholders involved in the clinical trial and happens in parallel with protocol development. QRM, starts with protocol definition, spans through study conduct till database lock. Quality is an extremely important aspect of a clinical trial, especially given the fact that it impacts patient safety. It is important for trial sponsors to have a robust risk management mechanism in place to ensure that patient safety is not compromised. A proactive approach in identifying risks in a clinical trial would help the sponsors to put in place an effective risk mitigation plan and avoid fatalities, budget overruns, major adverse events, etc. attributable to lack of oversight.
Traditional QRM in clinical trials focused mainly on site visits, in-situ source data verification, source data review and audits conducted during and after a trial. However Clinical Trials Transformation Initiative (CTTI) encourages the use of Quality by Design (QbD) in 'Risk Based Monitoring' (RBM) of clinical trials. Quality by design has been successfully executed in the field of manufacturing, including pharmaceutical manufacturing, but has still not been fully translated in the clinical trial space as it is 'expert driven' rather than 'process driven'. There are 4 elements to QbD, Plan, Do, Check and Act.
Most pharmaceutical majors have established QRM processes in place to identify risks and come up with the mitigation plans. Our viewpoint is that they can take a more proactive approach and manage the risks in a better way to prevent/ reduce their occurrences. Pharmaceutical companies have huge repositories of past trials, both successful and unsuccessful ones and with the help of this vast repository, predictive statistical models based approach could be adopted to identify and quantify the risks early in the game. Based on the risk predicted for a study, the sponsor/ study team could come up with the appropriate mitigation plan which can be used in risk based monitoring. Predictive approaches have been used successfully in retail industry to enhance customer experience.
As a first step towards developing the model, Key Risk Indicators (KRIs) need to be identified. Some of the risk indicators could be protocol deviations, amendments to the protocol, quality issues, budget overrun issues, etc. The next step would be to identify the important study attributes that might impact the risk indicators. Examples of study attributes include, therapeutic area, study phase, statistical design, the number of subjects involved in a trial, etc. This would then be followed by model building, using appropriate statistical and machine learning algorithms, which would identify the patterns in the study attributes that lead to the behavior of risk parameters. The final step would be to use the models to predict the risk of a study. The predicted risk of a study can be categorized into different classes, namely, high, medium, low using color codes. This would help the sponsors in proactively coming up with a better risk mitigation plan for any trial in the future. This can be employed on an ongoing basis.
Infosys believes that statistics based predictive approach could very well be applied and adopted in clinical trials for better patient safety, quality of trials. Infosys, with its vast experience in ETL and data analytics space, would be an ideal partner to sponsors of clinical trials in aiding their proactive risk management process of clinical trials. Predictive analytics combined with RBM will have a very high potential to transform the way clinical trials are carried out.