Standardization (of Rates)
STANDARDIZATION (OF RATES)
Standardization (or adjustment) of rates is used to enable the valid comparison of groups (e.g., those studied in different places or times) that differ regarding an important health determinant (most commonly age). Although often presented in epidemiologic textbooks as a separate technique, it is in fact a specific application of the general methods to control for confounding factors. As such, many of the issues related to confounding and methods used to adjust for confounding can be applied to standardization. Historically, the need for age standardization was recognized well before the general concept of confounding was formalized. It has it roots in the earliest epidemiological studies—the first known reference to age standardization appeared in a publication by F. G. P. Neison in 1844. The most familiar application is in the presentation of age-standardized mortality or cancer incidence rates to explore temporal trends.
Two major approaches to standardization have been used, direct and indirect. Direct standardization is used when the study population is large enough that age-specific rates within the population are stable. When the population is small (or the outcome is rare), the number of events observed can be small. In that circumstance, indirect standardization methods can be used to produce a standardized mortality rate (SMR) or a standardized incidence rate (SIR).
Direct standardization is commonly used in reports of vital statistics (e.g., mortality) or disease incidence trends (e.g., cancer incidence). Indirect standardization has a played a major role in studies of occupational disease or studies of place and time-limited environmental catastrophes. Indirect standardization was introduced as a tool before direct standardization (1844 vs. 1899).
The standard approach to explaining standardization involves the concepts of expected and "observed" counts. In direct standardization, one estimates the rate that would have been observed if
|Direct Age Standardization|
|Age Group||Number of cases||Number of months||Mortality rate||Reference population||Expected number deaths|
|SOURCE: Courtesy of author.|
the study population had had the same age structure as the reference group (e.g., the number of cases of disease that would be expected if the disease rates in the study population were applied to the reference population). In indirect standardization, one computes the number of cases of disease that would have been expected if the disease rates from the reference population had applied in the study population. Dividing the observed case count by the expected count yields the SMR. A more modern approach to standardization recognizes that these methods are computing weighted averages of the age-specific rates.
To perform a direct age standardization, one first has to select a reference population. This population is arbitrary, although conventionally one uses either the World Standard Population produced by the World Health Organization, or a census population count for the country in which the work is being conducted. Next, one computes the age-specific rates within the study group. Then, one multiplies these rates by the number of people in that age group in the reference population. These expected counts are summed and divided by the total population size of the reference population to yield the directly standardized rate. This is illustrated in the example shown in Table 1. The crude mortality rate is 63/1,000. Standardizing to the reference population gives an age-adjusted mortality rate of 79,540/1,800,000 = 44/1,000. The adjusted rate is lower than the crude rate is since the proportion of the reference population in the oldest age group (11%), which has the highest age-specific mortality rate, is only 50 percent of that found in the study population (22%). This adjusted rate can be directly compared to
|Indirect Age Standardization|
|Age Group||Number of cases||Number of deaths||Mortality rate in Reference population||Expected number deaths|
|SOURCE: Courtesy of author.|
adjusted rates from other years to detect trends in mortality.
Indirect standardization uses the reference population to provide age-specific rates. Within each age stratum, one multiplies the reference rate by the number of people in the study population to determine the number of cases that would have been expected if that were the rate in the study group. These expected numbers are added up across all age groups and divided into the observed number to yield the SMR. Values greater than 1 (or 100, as the SMR is commonly expressed multiplied by 100) indicate a higher mortality than expected. It is possible to compute an indirectly standardized rate, but this is much less common than SMR/SIRs. Unlike directly standardized rates, one can not compare SMRs across time or place. One can however, compare SMRs for different outcomes within the same study population. This is a significant limitation to the use of SMRs. In the example given in Table 2, the researcher observed 458 deaths. However, based on the age-specific rates in the reference population, only 221 deaths would have been expected, yielding an SMR of 2.07 (or 207) suggesting a higher mortality rate in the study population than in the reference population.
The use of standardized rates is controversial. Any summary measure can hide patterns that might have important public health implications. For example, with age standardization, one might fail to detect age-specific differences in risk across time or place. This might arise if a disease is displaying an increasing incidence due to a birth cohort effect (people at younger ages might have a
One of the biggest potential abuses of standardized rates is by health care planners who use the standardized rates to estimate demand for services. This is incorrect practice. The standardized rate reflects the number of new cases that would arise in a hypothetical population. The actual number of cases expected is given by the crude rate, which should always be employed in health care planning analyses.
(SEE ALSO: Rates; Rates: Adjusted; Rates: Age-Adjusted; Rates: Age-Specific)