← Back

U.S. Healthy People 2010 Criteria for Data Suppression

At-a-glance
Application Level Registry Users Customers Users
Practical Applications Basic Management
Steward
Public Health Professional
Link: http://www.cdc.gov/nchs/data/statnt/statnt24.pdf

Overview:
This document from the National Center for Health Statistics (NCHS) was written in July 2002. Healthy People 2010 sets health goals for the United States and monitors progress towards these goals by presenting data from a variety of data systems. The manual lists the reasons data are suppressed and comments on sample survey data systems and population-count systems with regard to these issues. This article is short and well organized and would be a good reference to a registry that is thinking about data release and wants to consider what other major data systems are doing with regard to data suppression.

The main reasons data are suppressed from sample survey systems are because of small numbers of cases or large relative standard errors (RSE) both of which can lead to statistical unreliability. For example, estimates from the Behavioral Risk Factor Surveillance System (BRFSS), a sample survey system, are suppressed if the denominator is based on fewer than 50 sample cases. The National Health and Nutrition Examination Survey (NHANES), also a sample survey system, suppresses results based on fewer than 30 sample events or a RSE greater than 30 percent.

Population-count systems are also concerned with small numbers of cases and additionally with confidentiality. Data from the HIV/AIDS Surveillance System (HIV-AIDS) are not suppressed due to statistical unreliability; however, they can be suppressed to protect case confidentiality. Data from health districts and other health areas are suppressed based on population size and are released in accordance with individual States data release policies. The National Vital Statistics System, which contains data from birth and death certificates, suppresses results based on fewer than 20 events because of statistical unreliability.