2
$\begingroup$

I have deaths by counties from 1980-2010 by age and sex. I want to calculate the SMR and then map the rates. My question is what standard population should I consider.

If the SMR is Yi (# deaths observed in i) / Ei (# deaths expected in i). Where E is calculated by multiplying the mortality rate of the standard population by the population by age in the exposed population (counties). My question is whether I should take as a standard, for example, the national death rate of 2010 or I have to take the national death rate of each year in the period (1980, 1981..2010).

Thank you. enter image description here

$\endgroup$

1 Answer 1

0
$\begingroup$

You would need to use the age distributions for each country and each year to calculate what the expected mortality would be that year, and then tally those expectations up to a grand total 1980-2010 to use as the denominator for SMR.

$\endgroup$
3
  • $\begingroup$ Thanks for the reply. But I still have the doubt about the standard population, if I have to choose a year as a standard (e.g. 1981 rate) or if it is estimated for each year. I edited my question with a part of the table as an example. $\endgroup$
    – Mael
    Dec 11, 2018 at 0:01
  • $\begingroup$ @Mael there is no standard population. The "standard" is a death rate for every age and year. With that, for any country and year and age distribution, you can predict the # of deaths. That's the denominator. To get the SMR from 1980 to 2010, you have to calculate the numerators and denominators for all 30 years, add them up to one grand numerator, and one grand denominator, and that's the SMR. $\endgroup$
    – AdamO
    Dec 11, 2018 at 19:56
  • $\begingroup$ @Mael using a standard population is how you would do indirect standardization which is a differen statistic. $\endgroup$
    – AdamO
    Dec 13, 2018 at 19:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.