Some of you (namely my parents and PhD advisor) know that I worked as a Senior Biostatistican last year for Janssen Pharmaceuticals. While I ended up leaving for my current postdoc position, I used and learned some useful skills that I thought might be valuable to share in case you’re a statistician curious about working in the pharmaceutical industry. Of course this is completely dependent on the area you end up working in and is just my experience supporting mostly Phase I-III clinical trials.

  1. Communication, communication, communication. In my opinion, the most valuable skill you can have as a statistican in pharma is communicating statistical ideas & results to (very smart) non-statistics collaborators and statistician leaders who aren’t working with data/modeling day-to-day. This often means communicating assumptions in non-statistical language, using 1-2 effective visualizations to summarise a large amount of simulation results, and being clear about what the findings and/or recommendations are.

  2. Running simulations in a high performance computing environment. I worked in the statistical modeling & methodology group, and thus my job was to mostly support trial design, rather than work as a trial statistician. A lot of the questions I tried to answer were of the form “What would happen to our power/trial timing if this assumption is not true?”. This often meant coding up a simulation that encoded the usual assumptions for a trial with a time-to-event endpoint, and then being able to run the simulation for different combinations of assumptions (i.e. the drug is even better than we think it is, non-proportional hazards, accrual not as fast as we assume, etc…). While this could be done on a local computer, being able to have a pipeline to run these simulations on a HPC allowed me to respond quickly to collaborators, which they really like. I learned these skills during my PhD, and I would highly recommend taking advantage of learning how to use an HPC environment if your university has one.

  3. Learn the statistical background of conducting interim analyses (see more here). On the statistics side, learning the methodology and math behind interim analyses was one of the most complicated topics that I worked with on a day-to-day basis that I did not learn about in grad school. While you don’t strictly have to know all of the details, I found it incredibly helpful to know what was going on behind the scenes, as I found it helped with both coding simulations and also thinking about simulation results that seemed off at first glance. It was also interesting to think about why interim analyses are much more common in industry than in academic clinical trials, and why you might choose the timing & number of interim analyses for a trial.

If you have any questions about my experience, please feel free to get in touch with me. I would also recommend talking to other people as well!