Last update:
28-Dec-2007

Standardization is a method of controlling the confounding introduced by different age distributions in two or more compared groups. It is a method of combining category-specific rates into a single summary value by taking a weighted average. It weights the category-specific rates using weights derived from a standard population. The weights, in effect, define the standard population. Although a single set of rates can be standardized, the main reason to standardize is to facilitate comparisons and therefore, usually two or more sets of rates are standardized. In order to compare rates for exposed and unexposed people, both groups would be standardized to the same standard. This form of standardization has long been referred to as "direct" standardization, to distinguish it from alternative approaches, the most popular of which is "indirect" standardization. Whereas "direct" standardization is a process involving the weighting of a set of observed category-specific rates according to a standard population, "indirect" standardization is a process in which the standard, instead of supplying the weighting distribution, supplies a standard set of rates, which are then weighted to the distribution of the population under study. The process is used to generate an "expected" rate or an expected number for the crude rate or total number of cases in the population. The comparison between the study population and the standard is generally presented as a standardized morbidity (or mortality) ratio, which is the ratio of the "observed" number of cases to an "expected" number.

Key words: Confounder, Confounding, Standard population, Standardization, Stratification.