Biotech Values

[DewDiligence]

DNDN:

Before Impact release last Apr09, Dew was adamant it would fail to be stat sig due to PSB (programmed survivor bias).

Please get your facts straight. What I argued was not that the IMPACT trial would necessarily fail, but rather that neglecting to take into account program-survival bias was causing some posters—including simulation modelers such as Ocyan—to be unduly bullish about the hazard ratio of the IMPACT study in the final analysis. If I recall correctly, Ocyan initially predicted a hazard ratio* of more than 1.50 and later reduced his prediction to something closer to 1.40. The actual hazard ratio of the IMPACT study in the primary analysis turned out to be 1.29, which was closer to my prediction than to Ocyan’s.

*Using the reciprocal method of presentation favored by DNDN.