Biomedical Engineering Reference
In-Depth Information
which diagnosis codes are used may also be inad-
vertently affected by coding practices designed to
simplify data entry, such as the use of super-bills or
the format of electronic input systems, e.g., having
pre-selected subsets of diagnosis codes available in
pick lists. For example, experience with the Depart-
ment of Defense's Electronic Surveillance System
for the Early Notification of Community-based
Epidemics (ESSENCE) system [11] has shown
that the most commonly used code during several
Norovirus outbreaks at military treatment facilities
is actually a non-infectious gastroenteritis diag-
nosis. Therefore it is recommended that developers
of new systems analyze the data of the institutions
to be included in the systems and adopt coding
strategies to accommodate those practices.
Another issue for consideration in adopting a
diagnosis-coding based syndrome scheme is the
sheer number of diagnoses included in a syndrome
category. Some diagnosis codes may be used so
infrequently that an important increase in their
frequency is lost among the background of other
more commonly used codes. It is therefore impor-
tant to examine the frequency of each code in
the syndrome group under normal circumstances
and to characterize the attributes of the codes as
a group in order to determine whether they will
be useful in detecting changes in disease inci-
dence. Additionally, when possible, it is valuable
to compare diagnosis code groups to a “gold stan-
dard” data source that exemplifies the trend for
the diseases the syndrome addresses. Figure 3.1
shows an example of a comparison designed to
determine which set of ICD-9-CM codes is best
able to capture the seasonal influenza trend seen
in positive viral respiratory specimens collected
in a laboratory-based surveillance system [12].
This analysis shows that manipulating the selec-
tion of diagnosis codes can produce more specific
syndromes that fit the needs of a particular surveil-
lance program, such as the influenza-like illness
(ILI) category used by the ESSENCE system.
Each individual hospital or other care facility
should analyze what information is captured in
their new patient encounters to determine what
information might be useful
for
surveillance
Viral Respiratory Specimens vs. ICD9 Codes
140
100000
90000
120
80000
100
70000
60000
80
50000
60
40000
30000
40
20000
20
10000
0
0
Weeks
Positive Specimens
Influenza-like Illness syndrome (ICD9 Codes)
Figure 3.1
Frequency of ICD-9-CM codes grouped as influenza-like illness compared to specimens positive for influenza virus.
(Specimen data was provided by the Air Force Institute for Operational Health.)
