Suppression of Incidence (and Death) Rates (and Trends) and Case Counts

When the numbers of cases (or deaths) used to compute the incidence (and death) rates (and trends) are small, those rates (and trends) tend to have poor reliability. Therefore, to discourage misinterpretation and misuse of counts, rates, and trends that are unstable because of the small number of cases or deaths, these statistics are not shown in tables and figures if the counts are less than 16 for the time period. A count of less than approximately 16 in a numerator results in a standard error of the rate that is approximately 25% or more as large as the rate itself. Equivalently, a count of less than approximately 16 results in the width of the 95% confidence interval around the rate being at least as large as the rate itself. These relationships were derived under the assumption of a Poisson process and with the standard population age distribution close to the observed population age distribution.

Another important reason for employing a cell suppression threshold value is to protect the confidentiality of patients whose data are included in a report by reducing or eliminating the risk of identity disclosure. The cell suppression threshold value of 16, which was selected to reduce misuse and misinterpretation of unstable rates and trends in this report, is more than sufficient to protect patient confidentiality given the low level of geographic and clinical detail provided.

United States Cancer Statistics Report (2004 Incidence)
page 10