 |
The impact of the
redesign of the NHIS has some general implications for disease surveillance.
First, disease prevalence estimates are a product of the case definition.
For the example of asthma, there are various case definitions based on the
source of data--administrative records, self-report, or clinical/laboratory
data. (See the 1998 Position Statement by the Council of State and
Territorial Epidemiologists (CSTE) at
http://www.cste.org/ps/1998/1998-eh-cd-01.htm). For each type of data,
case definitions are classified by decreasing certainty as "confirmed,"
"probable" and "possible." Note that there is no confirmed case definition
for asthma using self-reported data, because self report can be affected by
numerous factors including the knowledge of the reporting individual, the
accuracy of recall, clarity of the question, etc. The questions on the
redesigned NHIS better reflect the CSTE "probable" case definition for
self-reported asthma than the pre-1997 question. Of course, there also can
be major changes in a survey instrument that lie behind fluctuation in
trends. Most of the national data systems that collect health information
undergo periodic revision, such as changes in birth and death certificates,
updates in sampling design, incorporation of information about new health
technologies, and question changes. Question changes can reflect updates in
case definitions, or an effort to improve data collection independent of a
change in definition. It is important to become knowledgable about the
underlying data collection methods, any major changes to these methods, and
the possible impact of these changes when analyzing any data system. |
