2014.ISDS.Abstracts.Final.pdf


ISDS Annual Conference Proceedings 2014. This is an Open Access article distributed under the terms of the Creative Commons Attribution-
Noncommercial 3.0 Unported License (http://creativecommons.org/licenses/by-nc/3.0/), permitting all non-commercial use, distribution, 
and reproduction in any medium, provided the original work is properly cited.

ISDS 2014 Conference Abstracts

Tuberculosis Unseen - Missed Opportunities in 
Diagnosis

Aaron C. Miller*, Linnea A. Polgreen and Philip M. Polgreen

University of Iowa, Iowa City, IA, USA

Objective
To estimate the potential number of Tuberculosis (TB) cases 

that occur in inpatient and emergency department settings that are 
missed, diagnosed as something else, go untreated and return to 
the community, prior to receiving a correct diagnosis of TB. We 
analyze inpatient and emergency department records from the state 
of California from 2005-2011.

Introduction
Historically, patients with TB have often been diagnosed late1 

or after death2. This delay in diagnosis often occurs because TB 
is misdiagnosed as an alternative respiratory illness (RI), such as 
pneumonia1. TB infected patients that are not correctly diagnosed 
when initial symptoms occur may spread infection to others in both 
healthcare settings and the community.

Methods
HCUP state inpatient and emergency department records for the 

state of California were extracted for patients with multiple stays 
between 2005 and 2011. Records containing an ICD-9 diagnosis code 
beginning with 010, 011, 012, and 018 were used to identify patients 
with a TB diagnosis. Next, we searched for visits that preceded 
the patient’s TB diagnosis. Patient records following the first TB 
diagnosis and those containing a prior alternative TB diagnosis (e.g. 
ICD-9 codes starting with 015 – TB of the bones and joints) were 
excluded. We defined a potential missed diagnosis (PMD) to have 
occurred if: (1) a RI was indicated in the previous visit and (2) that 
visit occurred in an event window of 5-90 days directly prior to the 
admission containing the initial TB diagnosis. If the previous visit 
occurred less than 5 days prior to the TB associated visit, it was 
excluded because there is a time lag between TB testing and diagnosis.

Finally, in order to determine if the RI’s in a visit preceding a TB 
diagnosis represent a PMD rather than a correct but coincident RI 
diagnoses, the occurrence of such events was compared between 
patients with and without an initial TB diagnosis. Odds ratios were 
computed for the probability of a RI occurring in a previous stay 
falling in the same event window between patients with and without 
a TB diagnosis. Multivariate logistic regression was also used to 
estimate this odds ratio while controlling for patient characteristics.

Results
A total of 6,664 initial cases of TB were identified, 3,164 of which 

had a RI in the visit directly preceding their initial TB diagnosis. 
Of the initial TB cases identified, approximately 23.33% had a 
PMD, defined by a RI. The rate of PMDs increased with larger time 
windows: PDE rates were 19.75%, 25.33%, 28.45%, to 32.5%, when 
60, 120, 270 and 360 day time windows were used, respectively.

In comparison to the non-TB patients, those with an initial TB 
diagnosis were far more likely to experience a RI in a previous visit 
(OR 3.86, CI 3.58-4.16), and only 8.27% of the previous visits in 
the non-TB population contained a RI. After controlling for patient 
characteristics, the odds of an initial TB diagnosis having a previous 
visit containing a RI was 2.77 (CI 2.55-3.00) times the odds of a 
non-TB visit.

Conclusions
Our results indicated that misdiagnosis of TB may be occurring in 

more than 20% of TB patients diagnosed in inpatient and emergency 
department settings. While some of these events likely represent 
coincident RIs rather than PMDs, patients with an initial TB diagnosis 
were far more likely to experience a RI in the months leading up to 
diagnosis than non-TB patients. Moreover, given that RIs are often 
diagnosed and treated in outpatient settings, even more misdiagnoses 
may be occurring than can be observed in the data used here. These 
findings call for an increased awareness in TB surveillance, especially 
to prevent healthcare associated spread of TB.

Keywords
Tuberculosis; Misdiagnosis; Disease surveillance

References

1. Wang J, Hsueh P, Jan I, Liaw Y, Yang P, Luh K. Empirical treatment
with a fluoroquinolone delays the treatment for tuberculosis and is 
associated with a poor prognosis in endemic areas. Thorax. 2006; 
61(10)903-8.

2. Bobrowitz I. Active Tuberculosis Undiagnosed Until Autopsy. Am. J.
Med. 1982; 72(4): 650-8.

*Aaron C. Miller

E-mail: aaron-miller@uiowa.edu    

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * (1):e41, 201