We observe the event of death for a sample of individuals aged x (50 – 100 years) (Life insurance) during a year (t)
Objective : estimate the death rate at time (t) and age (x)
Data: Population and deaths
Population: number of individual alive (aged x) by the end the year (t) all generated from the same cohort (g).
Deaths: The number of deaths occurred at the age (x) and time (t). These deaths don’t originate from the same cohorts. We can’t distinguish death by cohort (g or g-1)
We aim to estimate the population at risk for age x and time t corresonding to the observed deaths.
The problem: Some individuals enter late to the sample (Insurance subscription)
Questions: How to estimate the specific age death rate ?
How to do it with R? (Survival package and KMsurv) or other methods
find a simplist shemes in attachment
Give a try at:
https://www.cscu.cornell.edu/news/statnews/stnews78.pdf
The problem is that we can't observe the date of entrance to the sample, and we observe only the surviving population by december 31th and the total number of deaths. also, we dont have any assumption about the anniversary dates (individuals born in january 1st are exposed 12/12, who born in September 4/12)
If you have no date of entrance, you can assume that the individuals have their average entrance at June 30. The Standard error should small, otherwise you should used delayed entry time points for each individual. Then you habe two time scales. First ist the total survival time from date of birth, to date of death. Second the time between date of birth and the date of entry.
thanks for your answear. That's what i did finally, For entrance and censured individuals, I assumed that they were exposed for the death risk only for 6 mounth during the observed year. The same is assumed for deaths, The dates of death were not currently observed too.
- Badges
- Science topic