Background information of SMR

The Standardized Mortality Ratio (SMR) is an internationally recognized health status indicator (British Columbia Vital Statistics).  It is defined as

SMR = (Oi / Ei)

Where Oi is the observed count and Ei is the expected count

 SMR is a relative measure between the difference of the exposed and unexposed populations within an area. For example, if observed mortality is very similar to the expected mortality, the observed over expected values will result in a value equal to one (1). SMR values greater than one represent areas with higher than expected levels of mortality. Likewise, SMR values less then one represent areas with lower than expected levels of mortality.

 The limitations of SMR

 The estimation of SMR is associated with the different level of variability since the expected counts are estimated based on the population size. As population size differs from place to place, there is less consistency with an varying denominator. The standard error of SMR is calculated by:

As the standard error of SMR increases the reliability of the SMR ratio decreases.

In addition, the situation that the observed number is zero is also concerned since the SMR in the area becomes zero, regardless of the size of the population (Meza, 2003).  If the study of interest is a rare disease, zero observation can be possible, but not necessarily beneficial as information can go unnoticed. Research methods that identify the empirical Bayes estimation adjust the standardized mortality ratio to overcome its limitations. This encompasses the smoothing of ratio values according to the overall mean of the SMR.