If I completed a study I'd certainly extract the good points in a NY minute. Why wouldn't you? The bears call it mining and I call it using the facts to your advantage.
True. And you would release the best data. Where's the good data? A 9% mortality benefit in a 62 patient slice which represents 16% of the total patients is likely due to randomness. If you flip a coin 393 times and then divide the results randomly into 6 groups, one of those groups will likely show a 9% benefit.
The term "data mining" is a tool shorts use to distract from positive results. You'll see it being used when trying to draw attention away from positive data.