It means the overall death rate in the sample group was decreased substantially. The number of people who survived because they didn’t get lung cancer or blood clots was so large that it had a noticeable impact on the number of total survivors, even when you include death by bus. This is a useful measure for a couple of reasons. One, it accounts for the prevalence of the disease being prevented - cutting all pork from your diet prevents 100% of deaths by trichinosis, which accounts for like 0.00001% of deaths from all causes (completely made up numbers and example, without consulting any sources). Two, it could account for net change in survival, for a treatment or behavior that has both positive and negative effects - giving radiation therapy indiscriminately to everyone with any kind of lump might decrease rate of dying from breast cancer, but increase death “from all causes” because it causes more problems than it solves.

I guess an additional way it might be useful is if we don’t yet have data on the exact mechanisms by which the treatment helps or what exactly its preventing - all we know is that we gave group A the treatment and not group B, and after 20 years there were a lot more people alive in group A, but we haven’t yet found a pattern in which causes of death were most affected and how.