RBQM Ops 2022 - Data Quality in Observational Research

Data Quality in Observational Research

DUNCAN HALL:

The next topic is How does statistics drive study goals and quality? Statistics are at the heart of all clinical research. So we really wanted to get a highly qualified statistician to take us through this important subject. I'm delighted to introduce Saverio Verdone, and the thrombosis Research Institute, a research fellow at the London School of Hygiene and Tropical Medicine. Saverio is a very well respected, well published author with over 50 Publications accredited and recommended by many statisticians from leading global organizations. So Saverio, thank you for joining us today and over to you.

SAVERIO VERDONE:

Thank you, Duncan. And thank you for attending this talk, where we're going to discuss why data quality is a salient issue, especially in the field of observational research. And we're going to see some approaches, some things we need to take into account, either when evaluating the quality of a published observational study, or when designing or analyzing the data of a new observational study.

So, if you're here, you probably already familiar with the main differences between randomized control trials and observational studies. But just to clarify, randomized control trials have randomization. Randomization creates groups that are being compared, balancing some baseline characteristics. And this is where randomized control trials are often considered the gold standard for estimating causal effects.

On the other hand, observational studies, they have numerous inherent biases. Findings of observational studies are often suspected of being skewed by all these biases. And this is why observational studies are often seen as unreliable, at best useful for generating hypotheses, and at worst responsible for false alarms and therefore not worthy of consideration.

Now, in reality, the degree of reliability of observational studies varies widely, in particular, according to the quality of the data recorded. So yeah, observational studies have issues. So why should we care then? Actually, because observational studies are a major and indispensable source of medical and scientific knowledge...