I did an internship with Eli Lilly last summer as a mechanical engineer. I had to do a standard 5-panel urine test. It had been pretty much exactly 30 days I had last smoked and everything went fine. I have since received a full-time job offer, so I've been abstaining from illicit substances for a couple weeks now.
It's possible that I will be asked to do a blood test for the full-time offer, but I don't know.
And yes, Lilly tests their drugs. The FDA requires three phases of clinical trials for any drugs by any company. Since Lilly mostly markets new drugs, they have to make all those millions (yes, millions. I know. It's a lot.) of dollars they spent on testing back before the patent runs out, and other companies can start selling the same drug. Since those companies didn't have to spend gobs of money developing the drug in the first place, they then sell it for cheaper.
Scientists and engineers at Lilly work hard, and have gone through total hell in school to have the privilege of doing what they do. Our biggest fear is that our products may end up hurting someone. Sometimes, there is statistical evidence of technological shortcomings, like with Cymbalta. This is very unfortunate for the company, the developers who worked so hard, and especially the patients.
Keep in mind that Lilly can't sell anyone anything unless doctors prescribe it. If a doctor, who has been though 15 years of total hell in school and knows far more about these matters than anyone who would spend time on this message board (yes, including myself), decides that he/she wants to perscribe Cymbalta, it is because he/she has decided that the benefits outweigh the risks.
Or you just might have a crappy doctor.
Lilly cannot make money selling drugs that make people want to do themselves in. The instances of the Cymbalta suicides are statistically up in the air. People who take this medication are depressed in the first place, so they already have a higher suicide rate. Cymbalta does not fix them, it makes them more functional in society. The statistical P-value for the hypothesis that Cymbalta induces or increases risk of suicide is too low to say there is any actual effect.
There is no "evil" company. Sometimes individuals within the company are just asshats. All companies want to make as much money as possible.
If you really want to cut into Lilly's profit margin, then don't buy oncology medication when you get cancer.