Summary: company funded studies claim drugs help X% of the population. More rigorous federal studies show they might only help 2/3 * X%. The main difference in the groups was including drug abusers in the federal group.
So what? A drug either helps you or it doesn't. If it helps you, you don't care what fraction of other people it helps.
The value of X mainly determines how many drugs you'll need to try. If they work for 50%, you'll have to try 2 drugs on average before finding one that works. So you want to start with drugs with large values of X, but whether it's 60% or 40% isn't a huge deal.
Furthermore, clever doctors can often guess right and prescribe different drugs for different people. Studies that insist on giving the same drug to everyone with depression are stupid, because they're administered much more intelligently in the real world.
So what? A drug either helps you or it doesn't. If it helps you, you don't care what fraction of other people it helps.
The value of X mainly determines how many drugs you'll need to try. If they work for 50%, you'll have to try 2 drugs on average before finding one that works. So you want to start with drugs with large values of X, but whether it's 60% or 40% isn't a huge deal.
Furthermore, clever doctors can often guess right and prescribe different drugs for different people. Studies that insist on giving the same drug to everyone with depression are stupid, because they're administered much more intelligently in the real world.