Certainly not all the time.
For the second time in 18 months, I have had a friend or acquaintance, approximately my age, die of cancer.
The man who just died? He was given six months to live after a diagnosis of terminal cancer. Didn't make it that long.
The woman who died in October 2012? She had a recurrence of cancer that hit so fast she was in a coma within 24 hours or so of being rushed to the hospital. (I suspect it appeared even faster than it was, because Kishi was trying not to tell her husband.) She died before that week was out.
The claim that many doctors are conservative on their diagnoses, usually on cancer cases, is often tied with a mislabeling of spontaneous remissions of cancer as "miracle cures."
Well, there are no such things.
What there is, is a whole constellation of diseases, many of them still poorly understood, lumped under the label of "cancer."
I'm not a doctor, so I don't know for sure all of the reasons doctors give diagnoses that often, but by no means always, turn out to be conservative.
One might be for the doctor wanting to appear heroic and salvific by underdiagnosing the likelihood of full recovery, rather than appearing "defeated" with too optimistic of an assessment.
Another might be a "by the book" system that says something like: "Stage IV bladder cancer, diagnosed one week ago, X percent advanced = Y months left to live."
In any case, I see "conservative" diagnoses and alleged "miracle recoveries" all tied together. Doctors are afraid to be fully human, patients and families put them on a pedestal, and, at the same time, magical thinking still permeates much of American life.