JSTOR

Methods

Newly hired HCP are screened by the Department of Occupational Health and complete a tuberculosis risk assessment questionnaire. The questionnaire includes information on history of BCG vaccination, country of birth, history of travel, tuberculosis exposure (by working in long‐term care facilities or prisons or from other possible contacts with tuberculosis patients), previous TST, previous chest radiograph, and previous therapy for positive TST result, if any. Beginning July 1, 2007, the TST was replaced by the QFT‐G In‐Tube test.

Phlebotomists at the Department of Occupational Health draw blood from HCP and place it directly into three 1‐mL tubes, each containing either (1) tuberculosis‐specific antigens, (2) phytohemagglutinin (a mitogen used as a control with a positive assay result), or (3) saline (which is used as a nil sample to measure the background level of IFN‐γ). Samples are to be vigorously mixed by hand after collection. The samples are then picked up by a courier and delivered to the laboratory, where they are accessioned and incubated at 37°C for 16–18 hours overnight. After overnight incubation, samples are processed and tested for quantitative IFN‐γ (IU/mL). Each test run is validated with external standards before the results are reported as positive, negative, or indeterminate.

Results

A total of 1,528 QFT‐G In‐Tube tests were performed in batches of 70 during the 4‐month period of our study (ie, from July through October 2007), and there were 1,292 (84.6%) negative, 68 (4.5%) positive, and 168 (11%) indeterminate test results. Of the 168 tests with indeterminate results, 167 (99.4%) were due to low mitogen (ie, phytohemagglutinin) response, with IFN‐γ at less than 0.5 IU/mL (with no obvious reason for the impaired cellular immune responses observed among the HCP). Only 1 employee had an indeterminate test result that was due to more than 10 IU/mL IFN‐γ in the nil tube.

Because the expected prevalence of indeterminate test results among HCP at the Cleveland Clinic is less than 5%,⁴ an investigation was performed. Internal and external reviews by the Cellular Immunology Laboratory and by the manufacturing company found that the blood samples were not being picked up and transported in a timely manner by the courier. The manufacturer recommends that tubes be incubated at 37°C as soon as possible and within 16 hours of collection. A change in the courier service did not affect the rate of indeterminate QFT‐G In‐Tube test results. A review of the different procedures involved in blood‐sample transport (ie, from blood collection through blood‐sample delivery to the laboratory) was undertaken. During observation of the phlebotomists at the Department of Occupational Heath, it was noted that winged blood collection sets were not used, which can make it difficult to control the amount of blood drawn into the tubes and which might have resulted in overfilling. The tubes were also inadequately shaken after blood collection. Antigens are dried onto the inner wall of the tubes, so it is essential that tubes are shaken vigorously for 5 seconds to ensure that the entire surface of the tube has been coated with the blood.

On the basis of the review, the following interventions were then initiated: (1) an in‐service refresher course for phlebotomists on the proper way to collect blood and (2) the use of a vortex mixer (Fisher Scientific) for the mixing of each tube in the laboratory immediately before incubation. Compared with the period during which a vortex mixer was not used (hereafter referred to as the prevortex period), there was a significant decrease in the rate of indeterminate test results during the period in which the vortex mixer was used (hereafter referred to as the vortex period): 168 (11%) of 1,528 test results were indeterminate during the prevortex period, and 61 (7.9%) of 773 test results were indeterminate during the vortex period. A 2‐tailed P value of .02 was considered significant (Figure). The third intervention was the use of a modified antigen tube that contained a third tuberculosis antigen. The use of the vortex mixer together with the modified antigen tube further decreased the rate of indeterminate test results: 51 (6.1%) of 834 test results were indeterminate during the vortex period with new tubes ( ; Figure).

(31 KB)

Figure.  Comparison of rates of indeterminate QuantiFERON‐TB Gold In‐Tube test results during routine preemployment screening for latent tuberculosis infection among healthcare personnel for 3 different intervention periods.

Open New Window

Discussion

An indeterminate test result does not provide useful information with regard to the likelihood of M. tuberculosis infection. The optimal follow‐up of persons with indeterminate test results has not been determined. The options are to administer another QFT‐G In‐Tube test with a newly obtained blood specimen, administer a TST, or do neither. However, an indeterminate QFT‐G In‐Tube test result is a meaningful result, because it reflects either impairment of the immune system and/or technical errors during the testing process.⁵ An indeterminate test result may be due to either low response to mitogen (IFN‐γ levels of less than 0.5 IU/mL) or high nil (control) values. A low mitogen response can be due to a low T‐lymphocyte count, reduced T‐lymphocyte activity, or the inability of T‐lymphocytes to produce IFN‐γ.⁵ This can occur in immunocompromised hosts, such as patients with human immunodeficiency virus infection, malignancy, or renal dysfunction.⁶ High nil values may be due to the presence of heterophile antibodies or to intrinsic IFN‐γ production.⁵ Other causes of indeterminate test results include improper specimen transport (delayed transport can affect lymphocyte viability) and/or improper specimen handling or storage, as well as other technical factors (eg, incomplete washing of an enzyme‐linked immunosorbent assay plate).⁵

At our institution, there was an unexpectedly high rate (11%) of indeterminate test results among HCP. Most of the HCP did not show evidence of immunosuppression. This prompted a review of the QFT‐G In‐Tube testing process; during this review, the problem with the procedures used by the phlebotomists was identified. There was also the possible overfilling of the blood collection tubes, which can result in the dilution of the IFN‐γ level; there should only be 1 mL of blood in each tube. The tubes were also inadequately shaken after blood collection. According to the standard procedure, after blood collection, the tubes should be shaken vigorously for 5 seconds to mix the content with the antigen‐coated inner surface of the tube.⁵ Because there were inconsistencies in the manner with which the tubes were shaken, we introduced another step in the blood collection process. After blood is collected, each tube is placed in a vortex mixer for 10 seconds. This intervention alone decreased the rate of indeterminate test results from 11% to 7.9%, with statistically significant prevortex and vortex periods ( ).

The introduction of an additional tuberculosis antigen (TB 7.7) also decreased the rate of indeterminate test results. Although there was no statistically significant difference ( ) in the rates of indeterminate test results between the vortex period without the additional antigen and the vortex period with the additional antigen, there was a statistically significant difference ( ) between the rate of indeterminate test results prior to any intervention and the rate after the implementation of the interventions (ie, the vortex period without the additional antigen and the vortex period with the additional antigen; Figure).

Notably, the prevalence of indeterminate test results among HCP is still greater than 5% after the implementation of these interventions. Mazurek et al.⁷ showed that there were varying rates of indeterminate QFT‐G test results (ie, from less than 1% to 21% of test results). However, there is a consistently low rate of indeterminate test results among patients who are otherwise healthy.⁵ In a study of M. tuberculosis infection among healthcare workers in rural India,⁴ there was only 1 indeterminate test result that resulted in a negative test result after repeat testing. Whether there are other interventions that can be used to reduce the rate indeterminate test results remains the focus of further investigation.

On the basis of the fact that 51 (6.1%) of 834 QFT‐G In‐Tube test results were indeterminate during the vortex period, we anticipate that 150 of the 2,500 new HCP hired annually at the Cleveland Clinic will have indeterminate QFT‐G In‐Tube test results. At our institution, we have a protocol for handling indeterminate test results. If a new employee has an indeterminate QFT‐G In‐Tube test result, the test is repeated in 6 weeks. If the test result is still indeterminate, the employee will be evaluated for the presence of immunosuppression and for evidence of active tuberculosis (by use of a chest radiograph). If there is evidence of either immunosuppression or active tuberculosis, the employee will be referred to the Department of Infectious Diseases.

Another important and interesting issue is the reversions and conversions of QFT‐G In‐Tube test results among HCP over time with serial testing.⁸ The reporting and interpretation of IFN‐γ levels may be useful for healthcare workers who have to make clinical decisions.

If a healthcare institution were to replace the TST with the QFT‐G In‐Tube test for the screening of latent tuberculosis infection, both the rates of indeterminate test results and the individuals tested would need to be actively monitored. Individuals with indeterminate test results would need to be evaluated for possible immunosuppression. The addition of manual vortexing of the tubes before overnight incubation and the use of recently modified tubes were associated with a significant decrease in the prevalence of indeterminate test results among HCP. In a tuberculosis screening program, tracking the prevalence of indeterminate test results after the implementation of QFT‐G In‐Tube testing is suggested as a quality control measure.