© 2008 American Thoracic Society doi: 10.1164/rccm.200802-340ED
Screening Contacts of TuberculosisOvercoming Obstacles with an Old Tool
Tuberculosis and Chest Service
University of Alabama at Birmingham Lung Health Center
Tuberculosis and Chest Unit
In this issue of the Journal (pp. 1041–1047), Aissa and coworkers provide a model for predicting the risk of infection among contacts of patients with tuberculosis (TB) in a suburb of Paris (1). Latent TB infection (LTBI) status among the contacts was established using the traditional tuberculin skin test (TST), despite high bacille Calmette-Guérin (BCG) vaccination coverage. Eight independent risk factors were identified, including index characteristics of cavity and sputum smear (
While IFN- To maximize the cost-effectiveness in detecting LTBI, the number of persons to be screened should be reduced by targeting those with the highest risk for infection and/or disease. With a predictive model involving only simple index and contact characteristics, Aissa and coworkers were able to reduce the number of contacts to be screened by 26% while maintaining a false-negative rate parallel to the estimated background LTBI prevalence of 8% (1). Such reduction, although modest, highlights the potential utility of further risk stratification even among a conventionally targeted group, in line with the findings of a previous study (7). Better characterization of infection and disease risks will therefore help to inform decisions in the targeted screening of LTBI. In the absence of a gold standard, much uncertainty remains regarding the actual sensitivity and specificity of the currently available diagnostic tools for LTBI (3, 4). Cross-reaction between Mycobacterium tuberculosis antigens with BCG antigens is a well-known limitation of TST (4). Despite the intrinsic difficulty in verifying vaccination status, BCG vaccination at infancy does not appear to affect the interpretation of TST significantly in adulthood (8, 9). However, revaccination was a common practice in many places until relatively recent reports cast serious doubt regarding its additional efficacy (10). Significant discrepancies were also found between the performance of TST and IGRAs among adults with previous BCG vaccinations (3). The specificity for TST is known to increase with higher cutoff points (4), but the trade-off in sensitivity can be substantial (4–6). Recommendations vary regarding whether a different cutoff should be used for interpretation of TST among BCG-vaccinated TB contacts. Various cutoff points (5, 10, and 15 mm) have been used in different areas (4, 5, 11). The model of Aissa and coworkers is heavily dependent on the validity of one of its underlying assumptions, that the optimal TST cutoff point is 15 mm for BCG-vaccinated individuals (11). Theoretically, a higher proportion of recent infection is likely to be found among contacts. Higher TST readings could possibly be found among those recently infected (5), or there could even be generally augmented TST response among previously vaccinated individuals. A false-negative rate parallel to the estimated background prevalence in Aissa and colleagues' study is reassuring. The model predictions were also consistent when used in another contact group. However, such information cannot provide independent validation of the assumptions on which the model was built. Final proof of the model must therefore await longitudinal data on disease risk. The risk of developing active TB is known to be much higher in the initial 2 years after infection (4, 12). This may be readily understandable if we assume there is biological variation in host vulnerability. More susceptible individuals are likely to be selected out, leaving the less vulnerable ones behind. Unfortunately, none of the currently available diagnostic tests are able to differentiate adequately between recent and remote infections (3, 4). The predictive model by Aissa and colleagues does not seem to help in this regard. Among the eight independent risk factors, the highest adjusted odds ratios were observed for age and birth in higher incidence countries, likely reflecting their effect on background prevalence of LTBI. A recent survey in the United States (1999–2000) showed the background rate of TB infection to vary from 1.8% in those born in the United States to 18.7% in foreign-born individuals (13). Therefore, the 8% of contacts with false-negative results in the model may not represent those 8% of individuals estimated to have remote infections. If increased age and birth in a higher incidence country were eliminated as predictive variables, the false-negative rate would increase and be above the estimate of the background rate of France. Nevertheless, it seems quite reasonable for the background rate of contacts to be higher than the population average, because the contacts tend to share the same environment, and presumably exposure risk, as their index cases. Again, longitudinal data on disease risk will be necessary to verify the various probability levels chosen for model calibration. Smoking has been associated with both TB infection and disease (1, 14). For the relationship between smoking and TB infection, most evidence is based on TST, but suggestive association was observed in a recent study using an IGRA (15). Smoking is often assumed to be associated with multiple behavioral and social factors that might predispose to TB infection. However, the excess risk of infection was seen predominantly among current smokers, with ex-smokers being much less affected (1, 14, 15). The total number of cigarette pack-years did not, by itself, increase the odds of a positive test response (14, 15). Thus, increased exposure alone may not be sufficient to account for these observations. We believe that, even though much is yet to be learned, in places where resources are limited and contact investigation is incomplete or not done using this model may be a reasonable approach. The model should help reduce costs and personnel burden, and should provide a pragmatic and effective form of contact investigation. FOOTNOTES Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
100 acid-fast bacilli per field), and contact characteristics of age, active smoking, household exposure at night, first-degree family relationship, birth in higher incidence country, and free health care. The sensitivity, specificity, and the positive predictive values of this model were 0.93, 0.34, and 0.36, respectively, similar to those reported in another population (
release assays (IGRAs) might offer some advantages over traditional TST in terms of specificity, absence of boosting, and requirement of only a single visit (
