Should Causality Assessment Be Simple?

The determination of a causal relation between a product and an event is a critical step during the assessment of SAEs during in clinical development. There is ample literature on the topic published over the last decades, including a recent and interesting review by Tarek A. Hammad (https://lnkd.in/gCsVjW44)

For PV professionals, causality is essentially a binary scale (related or not related). Causality (combined with expectedness, which is another topic) drives the reportability decision to health authorities and other entities (IRBs, investigational sites, etc.) and is a foundation for signal detection activities.

However, clinical development teams often prefer a more nuanced causality scale (probably, possibly, probably not, likely/unlikely, etc.) as they feel it may help in summarizing the safety profile in clinical study reports by providing more options for categorizing the safety profile. The benefit of this approach has to be demonstrated in practice for regulatory reporting or signal detection activities.

While there were many attempts to create a consensus around a more sophisticated scale (for example: https://lnkd.in/gytRVC9C ), there is still a lot of subjectivity and variability regarding the perception of the nuances between a probably related SAE and possibly related SAE, or between a probably not related versus an unlikely related SAE. Causality determination involves a mix of experience, knowledge, science, data interpretations, and also many assumptions. Therefore, a binary choice is simpler, easier and increases consistency.

Sometimes, other elements may influence a causality decision when they should not. For example, when an investigator determines an SAE is related without any scientific rationale other than the chronological coincidence: the sponsor may disagree but will hesitate to confront the important investigator from a large site. When it happens, it can generate lengthy debates and unnecessary SUSAR reporting, especially when the sponsor has limited exposure and experience in safety. This is when an independent and experienced PV function can stand its ground and push in the right direction. Patient Safety is paramount, and being conservative in PV is important, but performing an incorrect and uninformed causality assessment creates unnecessary “noise” in the system, as it was clearly expressed by the FDA multiple times.

Causality determination is not easy but should be always expressed with simple terminology (related or not related): clear thinking deserves to be expressed in simple terms. Experienced PV resources are helpful when important decisions regarding safety are made during clinical trials, and causality choices should be simple.