These are the first reported operational outcomes of program efforts to switch from the TST to a blood assay for M. tuberculosis
. In San Francisco, TB testing using the IGRA was feasible and acceptable under routine programmatic conditions in several different clinics. We tested 4143 patients by IGRA and obtained usable results on 3829 (92%). Indeterminate results were uncommon (2%). The PPD-based IGRA used during this time period has since been superseded by newer tests using Mycobacterium tuberculosis
-specific region of difference 1 (RD1) antigens (such as early secretory antigenic target 6 and culture filtrate protein 10) for T cell stimulation. Available evidence suggests that these RD1-based IGRA tests are more specific than the TST, and have similar or slightly higher sensitivity in most situations[13
] But the operational experiences we describe remain relevant to any public or occupational health program currently using the TST and considering use of a blood assay for M. tuberculosis
IGRA results were available to the TB program on 92% of persons tested, representing a likely improvement in TB testing outcomes compared to TST. This outcome compares favorably to the reported return rate for TST reading in similar patient populations in other cities, including 84% at a needle-exchange program in Baltimore, 64% at a public health program in Atlanta, and 47% in a street outreach program for drug users in Long Beach. [14
] Since results are available on most patients tested by IGRA, program resources may be directed to improving follow-up of patients with positive IGRA results. Ample room remains for follow-up improvement; among the 819 patients with positive results, subsequent medical and radiograph evaluations were completed within 30 days in 524 (64%). Collection of comparable outcomes for TST was not possible from program data in San Francisco, since no surveillance system existed for TST results.
Surveillance for new instances of infection with M. tuberculosis has been previously attempted in San Francisco, but failed due to the decentralized and subjective nature of the TST and to the large resource investment required to maintain surveillance system function in hundreds of testing locations throughout the city. Laboratory-based testing allowed us to obtain accurate data for surveillance of LTBI from these IGRA pilot sites, including both denominator data for patients tested and longitudinal data alerting SFDPH to instances of newly detected M. tuberculosis infection, including the occurrence of IGRA 'conversion', i.e. positive results in a patient with previously negative results. Furthermore, the use of a laboratory-based test improved information availability for community providers and likely reduced duplicate testing, as previous results were available via an electronic laboratory result reporting system accessible to many community clinics. In San Francisco, since March 2005 a newer version of the IGRA (QuantiFERON-TB Gold®, Cellestis International), which uses RD-1 antigens for leukocyte stimulation, has been available to all SFDPH clinics and key community organizations providing services to homeless and immigrant populations. Through wider use of this newer IGRA, we have essentially created a laboratory-based surveillance system for infection with M. tuberculosis, with reliable reporting of newly detected instances of infection with M. tuberculosis among key populations such as the homeless, IDUs, and new immigrants. This advance in TB surveillance capacity may allow SFDPH to more accurately target case finding and prevention activities in the future.
Phlebotomy for IGRA-based TB testing proved acceptable among a majority of patients at both a homeless clinic and a refugee clinic. Phlebotomy refusal was uncommon (8%), and phlebotomy failed in only 8% of patients with a high prevalence of intravenous drug use. At the homeless clinic, among patients tested by IGRA, a positive or negative result was obtained in 99%, which is greater than the proportion of patients who returned for TST reading at that particular site prior to IGRA implementation (88%).
IGRA implementation cost the health care system in San Francisco approximately $33.89 per patient tested, most of which (83%) was incurred by the laboratory. This cost assessment includes only measurable direct costs of phlebotomy, specimen transport, and laboratory specimen processing. Assuming all other costs remain the same, we have projected that the more-expensive RD1-based IGRA, QuantiFERON®
-TB Gold, will cost approximately $37.39 per patient tested. In comparison, the Center for Medicare Services has estimated the cost of the TST to be $9.79. [17
] Despite this lower per-test cost for TST, several factors such as differences in test specificity, patient time requirements, and the potential to lower the laboratory cost for IGRA through automation raise the possibility that the IGRA may be the more cost-effective test. Future cost-effectiveness analysis comparing IGRA and TST use in TB testing will be necessary, and should include potential cost savings from improved patient evaluation rates and avoidance of treatment of patients with false-positive TST results.
Several operational challenges were noted during IGRA implementation. Same-day processing for specimens necessitated early and often inconvenient cut-off times for specimen collection, as well as use of a courier service. Patients attending afternoon or evening clinics were forced to return another morning for IGRA testing or to have a TST. Future versions of the IGRA that might allow more flexible timing of specimen collection and transport would mitigate this challenge. The occasional occurrence of discordant IGRA and TST results proved time-consuming for patients and providers to manage. Programs implementing IGRA should establish guidelines to help providers manage such events.
Our operational evaluation has several limitations. We did not randomize patients or sites to TST or IGRA. A randomized study would yield a more accurate comparison of acceptability and testing outcomes. Our evaluation was not intended to evaluate IGRA concordance with TST results; studies addressing that question have been reported elsewhere[13
] In accordance with CDC guidelines we excluded many patient populations such as children and HIV-infected persons from IGRA, but operational outcomes may differ in those groups. Patients with positive IGRA results may not have responded to SFDPH attempts to contact them, and instead may have been evaluated by private providers. Hence we may have underestimated the proportion of patients who were tested for TB after receiving positive IGRA results. Our cost estimates are specific for the health care system in San Francisco, and should not be considered generalizable. We did not collect data about costs incurred by patients, which may significantly differ between IGRA and TST. We used phlebotomy refusal to substitute for specific questions to patients about TST or IGRA preference. In our evaluation we did not extend data collection to include completion of treatment for LTBI, which would be the optimal outcome measure for feasibility, but may vary widely by setting and patient population. Lastly, our evaluation included only the first-generation PPD-based IGRA, but the use of new antigens should not necessarily affect programmatic outcomes such as feasibility and acceptability, which are likely to be common among blood assays for M. tuberculosis