Abstract
Empirical fluoroquinolone use for ≥7 days delays antituberculosis treatment despite immediate mycobacteriology study http://ow.ly/L8fIm
To the Editor:
Despite its excellent activity against Mycobacterium tuberculosis [1], empirical fluoroquinolone (FQ) therapy has been associated with delayed initiation of appropriate treatment [2–4] and acquired resistance [5, 6] in tuberculosis (TB) patients presenting as community-acquired pneumonia (CAP). A previous study conducted in intensive care unit revealed immediate mycobacteriology study may possibly prevent the delay in anti-TB treatment [7]. However, this hypothesis has not been confirmed.
With a longitudinal follow-up of >99% of the residents of Taiwan (>22 million subjects) since 1996 [8], the National Health Insurance Research Database (NHIRD) provided very suitable research material for this study to investigate factors associated with delayed anti-TB treatment in patients with pulmonary TB initially presenting as pneumonia. The impact of empirical use of FQ and prescription of mycobacteriology studies on the first consultation for pneumonia was particularly emphasised.
From the NHIRD, a total of 81 081 adult patients (age ≥20 years) with pulmonary TB in the period 2004–2009 were identified. Among them, 16 683 (20.6%) had pneumonia within 6 months prior to anti-TB treatment [9] and were followed-up until December 31, 2010 or until lost to follow-up (cancelled health insurance prior to December 31, 2010). A pneumonia episode was considered if all of the following three elements were present: 1) compatible diagnosis (International Classification of Diseases, ninth revision, clinical modification: code 480–486 or 507); 2) prescription of antibiotics; and 3) chest radiography performed.
The 16 683 pulmonary TB patients had a mean±sd age of 67.4±18.1 years (range 20.0–105.5 years). Among them, 69.7% were male and 72.9% lived in an urban area (population density >1500 persons per km2). Mycobacterial culture and M. tuberculosis nucleic acid amplification test (MTB-NAA) became more frequently prescribed on the first consultation for pneumonia, from 48.3% and 0.2%, respectively, in 2004, to 56.8% and 4.9%, respectively, in 2009. Delays in anti-TB treatment from the first consultation for pneumonia gradually shortened, from 53.3 days in 2004 to 48.5 days in 2009.
Of the 16 683 patients, 2051 (12.3%) received a prescription of FQ for ≥7 days on the first consultation for pneumonia (FQ group) while the rest formed the non-FQ group. The prescribed FQ was levofloxacin in 908 (44.3%) patients, moxifloxacin in 716 (34.9%), ciprofloxacin in 402 (19.6%) and ofloxacin in 25 (1.2%). The clinical characteristics, including age, sex, severity of pneumonia (requiring emergency room visit or hospitalisation, intensive care, and mechanical ventilation), presence of comorbidity (as previously defined [10]), and income status, of the FQ and non-FQ groups were similar. However, in the FQ group, the first consultation for pneumonia was less likely to be at a local hospital (including local clinics and community hospitals) (34.5% versus 45.3%, p<0.001), more likely to be in an urban area (75.8% versus 72.5%, p=0.002), and had a higher frequency of mycobacterial culture (67.6% versus 50.3%, p<0.001) and MTB-NAA (4.0% versus 3.0%, p=0.008) prescribed on the first consultation. Delays in anti-TB treatment was significantly longer in the FQ group (63.0±47.1 versus. 49.4±47.6 days, p<0.001).
In multivariate linear regression analysis (table 1), prescribing FQ for ≥7 days on the first consultation for pneumonia was associated with a delay of 16.50 days (95% CI 14.39–18.61 days) in anti-TB treatment. Age, underlying comorbidities, low income, first consultation for pneumonia in an urban area and pneumonia requiring an emergency department visit or hospitalisation were also associated with delayed treatment. Immediate mycobacteriology study, including mycobacterial culture (−15.78 days, 95% CI −17.20– −14.35 days) and/or MTB-NAA (−13.78 days, 95% CI −17.81– −9.74 days) on the first consultation for pneumonia and later TB diagnosis year (−0.65 days, 95% CI −1.06– −0.23 days; per year increment) were associated with shorter delay.
Sensitivity analyses revealed that prescription of FQ for ≥7 days on the first consultation for pneumonia was associated with delayed anti-TB treatment among six subpopulations: 1) the 8780 cases with immediate mycobacteriology study (14.70 days, 95% CI 12.43–16.96 days); 2) the 7903 cases without immediate mycobacteriology study (19.54 days, 95% CI 15.47–23.61 days); 3) the 2730 outpatient cases (requiring neither an emergency department visit nor hospitalisation) without underlying comorbidity (7.69 days, 95% CI 3.14–12.23 days); 4) the 1227 outpatient cases without underlying comorbidity and with immediate mycobacteriology study (7.58 days, 95% CI 2.75–12.40 days); 5) the 6590 cases who had bacterial culture performed and had no bacterial pathogen isolated during pneumonia event (11.52 days, 95% CI 8.41–14.62 days); and 6) the 4092 cases with a duration of pneumonia >21 days (25.48 days, 95% CI 21.54–29.41 days). In sensitivity analysis (5), prescription of susceptibility test was used as a surrogate for the presence of a bacterial pathogen, as bacterial isolates that could possibly be responsible for pneumonia are routinely tested for susceptibility in clinical practice in Taiwan.
Using the NHIRD, a powerful tool to investigate the influence and interactions between diseases and treatments, we found that despite the fact that immediate mycobacteriology study, including mycobacterial culture and MTB-NAA, might significantly reduce delays in anti-TB treatment [7, 11], empirical FQ use still delays treatment in patients with immediate mycobacteriology study.
In patients presenting with pneumonia requiring hospitalisation, FQs have been recommended as first-line empirical antibiotic therapy due to their broad-spectrum microbial coverage and their proven effect of reducing the length of hospital stay, thus being more cost-effective compared to the combination therapy of β-lactams plus macrolides [12]. The major concern of empirical FQ use for pneumonia is the masking of TB, resulting in delayed diagnosis and treatment [2–4], and further dissemination of the disease. Thus, the best management for pneumonia patients may be the use of FQs under the premise that TB is still being considered and all possible intentions have been made for timely diagnosis.
The results of the present study support the hypothesis that immediate mycobacteriology study reduces the impact of TB masking due to empirical FQ use. However, the benefit is marginal, with only a 1.8-day reduction. Because only 38.4% of TB patients in Taiwan are smear-positive cases [13], for the majority of TB patients, the smear-negative cases, a clinical response after FQ use still significantly delays the initiation of anti-TB treatment until the results of mycobacterial culture become available. Furthermore, the time interval from inoculation of the respiratory specimen for mycobacterial culture to notification of a positive culture may be lengthened after FQ use because its anti-mycobacterial effect may compromise the culture efficiency.
As shown in the analysis, immediate mycobacterial culture and MTB-NAA are the only two factors that can possibly shorten the delay in anti-TB treatment by ∼2 weeks. Further studies are warranted to confirm these findings and to optimise the use of mycobacterial culture and MTB-NAA for the early detection of TB in patients initially presenting as pneumonia. This is especially true in TB-endemic areas, where M. tuberculosis maybe the causal microorganism of ≥10% of pneumonia cases [14].
This study has some limitations. First, results of laboratory and radiological studies are unavailable in the NHIRD. As such, the diagnosis of TB and pneumonia may not be 100% accurate. Second, patients may receive other antibiotics during follow-up consultations, resulting in misclassification and biasing the result toward the null hypothesis, causing underestimation of the impact of FQ use [15]. Third, the protective effect of immediate mycobacteriology study may be confounded by the indication, resulting in an overestimation of its benefits.
In conclusion, empirical FQ use for ≥7 days delays anti-TB treatment despite immediate mycobacteriology study.
Acknowledgements
The authors thank the National Health Research Institute of Taiwan for providing the NHIRD. Details of the computer code for statistical analyses are available from the corresponding author.
Footnotes
Support statement: This study was supported by the National Science Council of Taiwan (grant NSC-101-3114-Y-002-003) and Centers for Disease Control, Taiwan (grant MOHW103-CDC-C-114-112302). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Funding information for this article has been deposited with FundRef.
Conflict of interest: None declared.
- Received January 28, 2015.
- Accepted March 20, 2015.
- Copyright ©ERS 2015