Volume 2: No. 3, July 2005
Confidence Intervals for Estimates in the United States National Hospital
Discharge Survey, 1979–2000
This flow chart shows how the SAS program performs relative standard
error (RSE) retrieval and standard error (SE)
and confidence interval (CI) computation. Raw data available on CD-ROM from the National Center
for Health Statistics, including data from the National Hospital Discharge
Survey (excluding those for newborns), is entered into the system to produce
an annual total, which is then reformatted and renamed. The first part of
the program, depicted in a box on the left, transposes the 22 annual RSE parameter tables on the
NCHS-issued CD-ROM into two new parameter tables by SAS array programming.
The second part, the COMPURSE program, calls the parameters and calculates SEs and CIs for annual totals, multiple-year summaries, and average annual totals of multiple years of NHDS
data as output.
Figure 1. The process by which relative standard errors (RSEs),
standard errors (SEs), and confidence intervals (CIs) are calculated for National Hospital Discharge Survey
(NHDS) data (excluding those for newborns) using the COMPURSE program.
NCHS indicates National Center for Health Statistics.
Return to article