Skip to main content

Standardization

The objective of standardization is the compare the rates of a disease (or outcome) between population which differ in underlying characteristics (age, sex, race, etc) that may affect the overall rate of disease; In essence it's another way to account for confounding.

Standardization of Rates

The difference between crude rates and standardized rates is that the crude rates are calculated on the population under study, whereas standardized rates are based on particular characteristic(s) as standard.

If the rates are calculated based on the specific characteristic(s), they are called specific rates (eg. age specific mortality rates, or sex specific mortality rates).

CI of Standardized Rate

Variance:

image-1664204565101.png

Standardized Rate is a Weighted Average:

image-1664204578136.png

Variance for a Standardized Rate:

image-1664204587751.png

There are two approaches to Standardization of Rates:

Direct Standardization (Most Common)

image-1664201721254.png

wi are the frequency from "standard" population, ri are rates from testing population. Very similar to the "weighted average" method from the last chapter.

  • Stratum-specific rates (ri) from population under study
  • Strata distribution (wi) from a "standard" population
  • My rates, standard weights

There are many choices for the 'standard' population.

  • Internal Standard - one of the groups under study, total population of the study
  • External standard - outside, a larger group (ex. US population)

Indirect Standardization

Used when rates in your population are not estimated well due to small sample sizes. We need :

  1. Crude rate in population/sample under study
  2. Strata distribution of popuation/sample under study
  3. Stratum-specific rates (and crude rates) for "standard" population

image-1664201787604.png

wi are the frequency from testing population and ri are rates from "standard" population.

  • The stratum-specific rates in the study of interest are not reliable due to small sample size.
  • Stratum-specific weights (wi) from population under study
  • Strata distribution (ri) from "standard" population
  • My weights, standard rates

Results are often presented as a Standardized Mortality Ratio (SMR):

image-1664206858103.png

We can also calculate the standardized rate based on the standard population rate and teh target cohort-specific SMR:

s(r) = (Population Rate) x SMR