Exudative Pleural Effusion Phenotype by Three-Test Rule

advertisement
Exudative Pleural Effusion Phenotype by Three-Test Rule
Introduction:
Pleural effusions are the abnormal accumulation of an excess of fluid in the space between lung tissue
and the chest wall. Pleural effusions are often suspected in patients with respiratory complaints who
have characteristic physical exam findings such as dullness to percussion. Imaging such as chest
roentgenograms typically confirms the presence of the effusion. However, identification of the
presence of a pleural effusion by physical exam and imaging rarely provides details as to the
composition of the effusion and often a fluid sample is obtained by aspiration (a procedure called
thoracentesis). The sample is sent to the laboratory for testing that often includes chemistry, cell count,
microbiology and other assessment such as specific gravity. The laboratory testing of pleural effusions
typically occurs along a timeline, with quick tests such as chemistry analysis occurring first, and other
tests, such as culturing the sample for microbiology content, taking place over days. The initial
laboratory testing of pleural effusions often attempts to categorize the effusion into broad categories
such as exudate or transudate. Identifying an effusion as exudative using initial laboratory testing may
help guide management because exudative effusions are traditionally associated with conditions such as
infection (the definitive detection of infection may take days due to the delay associated with culturing
techniques). In order to facilitate management, there are several sets of criteria used to analyze the
initial results of pleural effusion testing and to categorize the sample as an exudate or not. The ThreeTest Rule is a quantitative set of the criteria used to assess the initial laboratory findings of a pleural
effusion sample and to classify the effusion as an exudate or not. Our phenotype seeks to implement
the Three-Test rule as a query that can be automated and used to screen coded laboratory data in a
health care data warehouse to identify patients with exudative pleural. In addition to benefits of
automation, this approach avoids reliance upon ICD codes, which lack specificity with respect to
exudative pleural effusions.
Methods:
The Three-Test Rule is a set of quantitative criteria that compares pleural effusion sample laboratory
results for lactate dehydrogenase, protein and cholesterol to threshold values. If the pleural effusion
sample results for any of these three tests exceed the threshold value, then the Three-Test Rule
classifies the effusion as an exudate. As an initial approach, we implemented the phenotype as an SQL
query in the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) database. MIMIC-II is
a database of intensive care unit patient data collected from 2001 to 2008 and is available to external
researchers in a SQL format. As outlined in the table below, we mapped the criteria of the Three-Test
Rule for exudative pleural effusions to LOINC codes, to MIMIC codes and created a SQL query to
implement the rule.
Laboratory Test
LOINC Code MIMIC Code
Cholesterol
9818-0
50237
Protein
2882-9
50241
Lactate Dehydrogenase
2530-4
50246
We coded the Three-Test Rule for exudative pleural effusion classification as a SQL query (PostgreSQL)
for use in the MIMIC-II dataset as follows:
(-- pleural fluid cholesterol criterion
select subject_id
from labevents
where itemid = '50237'
and valuenum > 45
group by subject_id
union
-- pleural fluid protein criterion
select subject_id
from labevents
where itemid = '50241'
and valuenum > 112.5
group by subject_id
union
-- pleural fluid lactate dehydrogenase criterion
select subject_id
from labevents
where itemid = '50246'
and valuenum > 2.9
group by subject_id)
intersect
(-- age criterion to select patients above 2 years old
select ad.subject_id
from admissions ad, d_patients dp
where ad.subject_id = dp.subject_id
and months_between(ad.admit_dt, dp.dob)/12 > 2)
Results:
We successfully used the SQL query implementing the Three-Test Rule for classification of exudative
pleural effusions to screen the MIMIC-II dataset for a cohort of patients meeting criteria.
Discussion:
The clinical utility of classification systems for exudative pleural effusions is based partly on timing.
When a patient undergoes thoracentesis to obtain a fluid sample of the pleural effusion, the laboratory
tests needed to use the Three-Test Rule can be done quickly, and the classification of an exudate can be
used to suggest the possibility of conditions that may require urgent treatment, such as empyema
(infection). The use of criteria such as the Three-Test rule allows the patient to start treatment with a
presumptive diagnosis based on clinical conditions and the results of the Three-Test Rule, rather than
wait days for more definitive results (such as microbiology cultures) to become available. Because the
MMIIC-II dataset has additional information, such as whether or not an infection was detected in the
fluid sample of the pleural effusion, we hope to enhance our work with this phenotype by comparing
the classification based on an SQL query implementation of the Three-Test Rule with a comparison to
the available follow up test results (such as microbiology culture results).
References:
1. Saeed M, et. al., Multiparameter intelligent monitoring in intensive care II (MIMIC-II): A publicaccess ICU database. Critical Care Medicine 2011;39(5):952-960
2. Heffner JE, Diagnostic evaluation of a pleural effusion in adults: Initial testing. In: UpToDate, Sahn
SA. UpToDate. Waltham, MA (Accessed on March 2014)
John Kimbrough
Vojtech Huser
Download