But, given the life-safety issues of pharmaceutical projects, I thought this was a pretty good list:
risks (and benefits) associated with a product or with the process
used to develop, manufacture, and distribute the product. The following
questions should be asked at each stage of the product’s life cycle:
• What are the safety risks?
• Who is at the highest risk?
• What populations are at risk?
• Are the risks predictable?
• Are the risks preventable?
1. Select well-defined, validated metrics.
2. Use at least two different evaluation methods for key RMP goals
or objectives. Preferably, the different evaluation methods would be
both quantitative and representative to offset the biases that are intrinsic
to any single evaluation process.
3. Use qualitative data collected from a large and diverse group of
patients when quantitative data are either not available or not applicable
to the evaluation measurement. Qualitative data such as focus
group testing may be useful in assessing the effectiveness of education
and comprehension about safety and risk information.
4. Consider using evaluation methods to assess if each RMP tool is
performing as intended.
Now, it's point #2 that's worth a pause. There's no end of project management references to cognitive biases, the most compelling work done by Daniel Kahneman and Amos Tversky. And, there's wide acceptance of various estimating methods that use multiple estimators, like the Delphi method and its agile counterpart: planning poker. And, we all know the danger of the single point estimate: any single point is likely to be wrong, so better to estimate with multiple points in a range.
But, there's little literature that I've seen that recommends outright multiple methods to neutralize methodology bias: that's more aggressive than multiple evaluators. Multiple methods means approach the problem differently, and independently, and then see if a consensus can be reached on the risk.
This is sort of the RMP equivalent of having multiple contract designers work on a common requirement, each with their own methodology; no two independent designs will incorporate the same errors. So, when the two designs are applied against a common problem, it's not likely they will both experience an error at the same time. One or the other will always be successful.
It's conservative, it's expensive in the short run, but in the pharmaceutical domain, the long-term consequences of being wrong are too calamitous to put aside a short term expense.
Here's a nice lesson in risk and experiment evaluation using a pharmaceutical example from the Kahnacadey.org
Are you on LinkedIn? Share this article with your network by clicking on the link.