Biomedical Engineering Reference
In-Depth Information
of the patient encounter and certain security provi-
sions are followed, it is reasonable and prudent
to develop disease detection systems that allow
hospitals to monitor their patient populations for
important changes that could signal a deliberate
exposure to pathogens.
In addition to monitoring counts of syndromic
groupings of clinical presentations and diagnoses,
these systems also collect non-clinical health-
related data such as retail sales of over-the-counter
pharmaceutical sales or absentee rates of schools
or large employers in an effort to corroborate find-
ings from the clinical data. For the purposes of
this discussion, particular emphasis will be placed
on the hospital-derived data that may be useful to
hospitals and public health organizations as well
as the statistical methods that may be employed to
detect changes in health status.
Table 3.1
Representative ICD-9-CM Code Syndrome
Assignment
ICD-9
Code
Diagnosis
Syndrome
Subgroup
001.9
Cholera NOS
Gastrointestinal Cholera
003.0 Gastroenteritis,
salmonella
Gastrointestinal Salmonella
004.9
Shigellosis NOS
Gastrointestinal Shigella
005.0
Gastrointestinal Food
poisoning
007.5 Cyclosporiasis Gastrointestinal Cyclospora
008.00 Enteritis, E. coli NOS Gastrointestinal E coli
787.01 Nausea with vomiting Gastrointestinal Gastritis
787.02 Nausea alone
Poisoning, food,
staphylococcal
Gastrointestinal Gastritis
787.03 Vomiting alone
Gastrointestinal Gastritis
787.91 Diarrhea NOS
Gastrointestinal Enteritis
382.9 Otitis media NOS
Respiratory
Otitis Media
460
Nasopharyngitis,
acute
Respiratory
URI
462
Pharyngitis, acute
Respiratory
URI
463
Tonsillitis, acute
Respiratory
URI
486
Pneumonia, organism
NOS
Respiratory
Pneumonia
3.1 Hospital Inpatient Data
Hospitals routinely collect patient information that
could be useful for the detection of an institution-
specific or a community-wide health emergency.
Diagnosis codes, such as the International Code
of Diseases (ICD) Clinical Modifications, have
been used both domestically and internationally
to monitor the health of communities, particularly
for infectious disease outbreaks [4-9]. Diagnosis
codes are usually entered electronically into billing
databases and other information resources making
them a readily available source for data mining.
When assigned early in the patient encounter, diag-
nosis codes can expedite timely ascertainment of
health events. In many hospitals, however, diag-
nosis codes may not be assigned until the patient
is discharged from the hospital, thus limiting
their utility for outbreak detection. However, some
hospitals do record diagnosis codes early in the
patient encounter. To begin designing diagnosis
code-based syndromic surveillance systems, it is
first necessary to select the syndromes to be moni-
tored and to develop clinical definitions for the
syndromes according to the diseases whose presen-
tations they are intended to represent. Recommen-
dations for syndrome groupings have been put forth
by a work group that represented federal and local
487.8
Influenza w/
manifestation NEC
Respiratory
Influenza
490
Bronchitis NOS
Respiratory
Bronchitis
784.1
Pain, throat
Respiratory
Sore throat
786.2
Cough
Respiratory
Cough
public health agencies as well as academia [10].
These syndromes represent broad groupings such
as gastrointestinal, respiratory, and botulism-like
complaints or illnesses. Table 3.1 shows represen-
tative assignments of diagnosis codes to syndromes
and the potential to break syndromes into smaller
subgroups for better understanding of the types of
illness driving observed increases.
In developing syndrome coding rules, codes can
be initially assigned to syndromic categories on the
basis of clinical judgment. However, many coding
assignments reflect the lack of firm diagnosis, and
codes for symptoms such as “cough,” “diarrhea,”
or “viral syndrome” are commonly used when the
complaint or diagnosis is unclear or not completely
ascertained. Moreover, variability in coding prac-
tices exists among providers; between hospitals
serving primarily pediatric versus adult popula-
tions; and between inpatient, outpatient, and emer-
gency/urgent care facilities. The frequency with
