Will the FDA refuse approval in the face of the favourable ODAC 9 versus 1 vote?

I have recently purchased a ticket securing my seat on the MSB rocket ship. The company is compelling; both with respect to its leaders and its products. However, these, by themselves, are not always enough. My final question to myself as I was lining up at the MSB ticket booth was: "what does the ODAC 9 versus 1 vote really mean?". Here is what I found in a number of articles (by way of purely statistical analysis):

Association Between Food and Drug Administration Advisory Committee Recommendations and Agency Actions, 2008-2015 (Audrey Zhang, Jason Schwartz, Joseph Ross) [published in the Milbank Quarterly, 14 July 2019]

The authors looked at 376 recommendations made by the Advisory Committees between 2008 and 2015. The FDA agreed with 293 of these recommendations (77.9%). So, right off the bat, statistically speaking, we are hoping that we fall within the 77.9% of cases where the FDA agrees with the recommendation (vote). Pretty good odds if you ask me.

There were 83 recommendations made which the FDA disagreed with (22.1%). Of these, there were 62 instances where the FDA refused approval despite the Advisory Committee voting for approval - that is, a 16.5% probability. This is MSB's risk. Interestingly, however, 29 (of these 62) went on to win approval at a later date.

The research noted that "discordance was associated with lower degrees of advisory committee consensus and was more likely for agency actions focused on medical product safety than for novel approaches or supplemental indications" (my emphasis added). In MSB's favour, obviously, is the 9 to 1 vote (not unanimous but close) and the fact that their product is a novel approach.

The research also noted that almost half the time the FDA went against the 62 "approved" Advisory Committee recommendations was in the early years of 2008 to 2009. Less than 10% of the disagreement happened in 2014 to 2015. This seems to indicate that the FDA grows increasingly reluctant to go against recommendations over the passage of time.

AdComm Recommendations - How Often FDA Does Not Follow Them? (Mark Senak) [16 August 2016]

The author looked at 136 recommendations made by the Advisory Committee meetings between 2011 and the date of the article (being 16 August 2016). The Advisory Committee voted to approve on 95 instances. Relevantly, the FDA ultimately approved on 89 of these instances (93.7%).

FDA Advisory Committee Outcomes (Philip Ma, Navjot Singh, Jeff Smith and Seth Townsend) [October 2013]

The authors focused on 63 recommendations between 2001 and 2010 in which there was a "voting on approval question" for original new drug applications and biologics licence applications. In these meetings, committee members were asked to vote for or against approval of the drug of interest. The Advisory Committee voted to approve on 49 instances. Relevantly, the FDA ultimately approved on 43 of these instances (87.8%).

Warning Biotech Investors: FDA Advisory Panels' No Means No, Yes Means Maybe (Michael Becker) [7 February 2011]

The author looked at recommendations from meetings conducted during 2010 and 2011. The Advisory Committee voted to approve on 10 instances. Relevantly, the FDA ultimately approved on 5 of these instances (50%). The author helpfully noted the 5 instances where FDA approval was not granted including the voting numbers:



Conclusion

I was unable to find research detailing numbers relating to recommendations and approvals from 2016 to present. This may be a limitation to the statistics revealed below.

The 2019 publication suggests that there is a 77.9% probability of the FDA agreeing with the vote of an Advisory Committee. There is, however, also a 16.5% probability of the FDA refusing approval in the face of a positive Advisory Committee vote.

The 2011 publication suggests that on occasions where an Advisory Committee casts a positive vote of approval, there is a 50% probability of the FDA ultimately following this recommendation and approving. This number is a worry. However, I do note the very small sample size (only two years) and I also note the comment in the Milbank Quarterly which suggests that disagreements are becoming increasingly uncommon since at least 2014 to 2015.

Further, in answer to the paragraph above, I am persuaded by the (more recent) 2016 and 2013 publications which suggest that on occasions where an Advisory Committee casts a positive vote of approval, there is between a 87.8% to 93.7% probability of the FDA ultimately following this recommendation and approving.

Buckle up. My opinion only. Do your own research.