Death rate as a proxy for median lifespan

The mean age at death is the sum of ages-at-death divided by the number of subjects. Using the mean requires you to wait until the last one dies, which is alright for worms who live for weeks, but a human may live decades, so progress of these tests will be very slow.

Using the median -- half live longer and half less -- will speed things up. (The median of this series is 9 -- 6,7,8,9,11,12,14.) A human at 80 has a life expectancy of 8 years, so using this measure is better than waiting decades. But it still requires waiting 9 years.

Using a death rate and comparing it to a standard one (or that of a control group) may give results in just a couple of years. (Wait the first year for the treatment to have an effect and measure the death rate in the second. The standard equation, a smooth line,  lets you calculate the median. Note that the standard death rate doubles for humans in about 7 years.) But does the rate for test subjects follow the normal curve? For example, all the test subjects might drop dead in the third year, or they might have a low rate in the second year but rise after that to the control rate (regression to the mean). How can we find out?

There have been many tests done on worms, fruit flies, mice and people, using many drugs. Are the death rates for most of these "standard" or "non-standard"? Has there been a study? If they are all standard we can assume a new drug test is standard and see whether it's doing any good rather than waiting for everyone to die (for calculating the mean) or half the tested to die (for the median). If we can measure the death rate in two years instead of waiting 9 or more for a median we can save decades on human tests. So – have these analyses been done? Do the death rates for most of the tests follow the curve? Can you refer me to a study?

Alan Mole