Sometimes companies stack the odds in their favor by using improperly designed studies. A favorite tactic is to compare the new drug to a placebo when there's already a good alternative treatment available. Another ploy is to compare a correct dose of the experimental drug to the wrong dose of the control drug--so that the patients in the control group are under-treated or plagued by side effects.
However, Goldacre says, in general industry sponsored trials are better designed than independently sponsored trials. (He doesn't explain why that might be.)
The real problem is that the industry simply hides negative results. We know this is going on because pharmaceutical companies have been caught doing it, and because statistical analysis of the range of studies that do get published points to massive as-yet-undiscovered data suppression by industry.