QuantiFERON-TB Gold Plus Testing for the Detection of LTBI in Health Care Workers in the Kyrgyz Republic

Background: Health care workers (HCW) are at increased risk of TB infection due to their close contact with infected patients with active TB. The objectives of the study were (1) to assess the prevalence of LTBI among HCW in major TB hospitals of the Northern Kyrgyz Republic, and (2) to determine the association of LTBI with certain job positions, hospital units, or hospitals. Methods: HCWs from 4 TB hospitals in the Northern Kyrgyz Republic were tested with the interferon-gamma release assay (IGRA) Quantiferon-TB Gold plus (QFT) for the detection of an immune response to TB as marker of TB infection. Age was controlled for as a confounder. Univariate and multivariable analysis were performed using logistic regression to assess the association of the risk factors (age, job position, and hospital unit) with having a QTF positive result. Firth’s penalized-likelihood estimates were used for determining the odd’s of QFT positivity to account for the small-sample size. Pairwise comparisons using the Bonferroni correction (conservative) and comparisons without adjusting for multiple comparisons (unadjusted) were used to identify the categories where differences occurred. Results: QFT yielded valid results for 404 HCW, with 189 (47%) that had a positive test. In the National Tuberculosis Center, there was an increased odds to have a positive QFT test for the position of physician (OR=8.7, 95%, CI=1.2-60.5, p=0.03) and laboratory staff (OR=19.8, 95% CI=2.9-135.4, p<0.01) when administration staff was used as the baseline and controlling for age. No job position was found to have a signicantly higher odds of positivity following a pairwise analysis controlling for multiple comparisons. Fifteen of the 49 HCW with follow-up tests converted from negative to positive at follow-up testing. Conclusions: This is the rst report on prevalence and risk factors of LTBI for HCW in the Kyrgyz republic, and results indicate an increased risk for LTBI among physicians and laboratory personnel. follow-up Fisher’s exact test and logistic regression were used to assess associations of study sites and age, time between tests, or job position, respectively with conversion or reversion of follow-up tests. Signicant associations were those that had a p-value <0.05 for all analysis.


Background
Tuberculosis (TB) affects over 10 million people worldwide and it is estimated that nearly 1.8 billion people are latently infected with TB(1,2). Latent TB infection (LTBI) occurs when an individual develops a measurable immune response to TB bacteria, but symptoms and signs of clinical disease have not yet manifested (2). It is estimated that about 5-15% of individuals with LTBI will progress to active TB at a certain time in their life, and those with active pulmonary TB can transmit infection to up to 15 people (2). The land-locked Kyrgyz Republic in Central Asia is not only considered a high incidence country (2017 incidence 144/100,000; 7,695 TB cases noti ed), but also has a high burden of drug-resistant (DR)-TB (1,372 [43,5%] bacteriologically con rmed MDR/RR-TB among 3,151 bacteriologically con rmed cases in 2017) (3) increasing the importance of effective control measures. Controlling LTBI is considered a key to control the TB epidemic in a country(1,2). Latent TB infection can be detected by indirect immunological responses using methods such as the Tuberculin Skin Test (TST), or interferon-gamma release assays (IGRA) such as the QuantiFERON-TB Gold plus (QFT; Qiagen, Hilden, Germany). In countries where BCG vaccination (in Kyrgyzstan national BCG coverage from 2009-2019 was between 96-99%(4)) and interference with the TST is common, IGRA have a markedly higher speci city to detect and monitor LTBI (5,6) .
Health care workers (HCW) are at increased risk of TB infection due to their close contact with TB patients, particularly when performing high risk procedures such as cleaning, sample collection, sample testing, intubation, or resuscitation (7,8). Although effective training and TB infection control measures should reduce transmission of TB in medical facilities (9), the annual incidence of LTBI among HCW in high prevalence countries is estimated to be as high as 7.2% (10), which is greatly increased over the general public (8,10,11). The high occupational risk to HCW is a problem not only for the individuals themselves, but also to the general public as progression to active TB leads to transmission not only in the community, but also in the hospitals during interactions with patients (11). In order to develop and implement effective intervention strategies, it is important to determine and understand the actual prevalence and risk of LTBI among HCW in different job positions and departments within hospitals (12). Additionally, there may be differences between hospitals depending on the effectiveness of institutional infection control measures. Therefore, it is important not only to monitor LTBI infections among HCW, but also to identify areas of increased risk of infection.
Our study was part of the USAID funded Stop TB Transmission in Hospitals (StopTTH) project under the Central Asian USAID-Defeat TB initiative. It aimed to (1) assess the prevalence of LTBI among HCW in the major hospitals in the Northern Kyrgyz Republic using QFT, and (2) determine whether speci c hospitals, hospital units, or job positions are associated with increased risk of LTBI.

Methods
This study was conducted in the four major TB inpatient facilities of the three most Northern regions of the Kyrgyz Republic. Those were the TB hospitals of the National Tuberculosis Center (NTC) in Bishkek, Karabalta, Kemin and Karakol where 351, 104, 45, and 50 staff, respectively, were working in 2018. The NTC is larger than the 3 other hospitals and constitutes the central TB hospital of the country where advanced and complicated cases can be treated. The other 3 hospitals are rural and of simpler infrastructure.

Participants
All individuals who worked in the TB hospitals of Karakol, Kemin, Karabalta, and at the National Tuberculosis Center (NTC) in Bishkek were invited to participate in this study during their regular occupational-medical health examination. Participants completed a questionnaire indicating their age, job position, and hospital unit in which they worked. All participants had to declare their consent to give blood for IGRA testing, and to use their data including their QFT results for this study. All employees of the hospitals who gave their informed consents were eligible to be enrolled in this study. Employees under QuantiFERON-TB Gold plus Participants provided 4.0 ml of heparine-blood and all assays were run within 3 days of collection. As per the manufacturer's instructions (QuantiFERON-TB Gold plus®, Qiagen, USA), 1.0 ml blood was distributed among each of the 4 QFT tubes for negative control (nil), positive control (mitogen), TB1 and TB2 antigens. Results were categorized as positive, negative, or indeterminate as per manufacturer speci cations. Blood samples were transported to the respective hospital laboratory in less than 3 hours before start of incubation (36±1°C for 16 to 24 hours). Interferon-gamma concentrations were determined by means of the QuantiFERON-TB Gold plus ELISA kits using the automated microplate washer ELx50/8 and reader ELx800 (Qiagen-BioTek, Hilden, Germany). A positive result was assigned when TB1 or TB2 minus the corresponding nil-values were ≥1.0 IU/ml and >25% of the nil-value; a borderline positive result was assigned when both TB1 and TB2 minus nil were below 1.0 IU/ml but at least one (TB1 or TB2 minus nil) was above 0.35 IU/ml and >25% of nil. When mitogen was ≥0.5 IU/ml, nil ≤8.0 IU/ml, both TB1 and TB2 minus nil <0.35 IU/ml or <25% of nil, a negative QFT was assigned. In all other cases the QFT result was called indeterminate.
All participants at the NTC or Karakol were offered up to three follow-up QFT tests performed during the study period (February 2018-June 2019). The rst two valid tests taken at least one month apart were used to determine the QFT "status" of the HCW. A positive status (P) was assigned when both tests were positive, a negative status (N) when both tests were negative, a "converter" status (C) when the rst test was negative followed by a positive one, and a "reverter" status (R) when the rst test was positive followed by a negative one. The stability of respective status was determined when a third QFT test was performed within one month of the second. A stable status was assigned when the third test matches the second test, whereas an instable status is assigned when the third test does not match the second.

Statistical analysis
All statistical analysis was performed using STATA 16.0 (StataCorp LP, College Station, TX). Univariate and multivariable analysis were performed using logistic regression to assess the association of the risk factors (age, job position, and hospital unit) with having a QTF positive result. Age was controlled for as a confounder, and separate estimates were made for either job position or hospital unit as the risk factor for the outcome of QFT positivity. Administrative staff were used as the baseline as they best represent typical population in the Kyrgyz republic due to limited/minimal contact with TB patients. Due to the "rare event" of TB being detected in the baseline group, Firth's penalized-likelihood estimates were used for determining the odd's of QFT positivity to account for the small-sample size bias that is introduced. Pairwise comparisons using the Bonferroni correction (conservative) and comparisons without adjusting for multiple comparisons (unadjusted) were used to identify the categories where differences occur. Con dence intervals were used to determine signi cance for pairwise comparisons using the Bonferroni correction. Differences between the 4 hospitals were analysed using a mixed effect logistic regression, with site as a random effect, to account for any innate differences that were not measured. Additionally, the NTC was compared against the three other sites combined (Karabalta+Karakol+Kemin) for associations with risk factors; and a students t-test was used to compare the average age between the NTC and the combined other sites. Department differences were only looked for all 4 hospitals combined due to decreased sample size, and similarities in department categorization between hospitals.
For follow-up analyses, the second QFT test result was used unless there was less than one month that had elapsed between tests of an individual HCW. In this case, the 3 rd test was used if available and more than one month had elapsed. Participants with less than two valid QFT taken at least one month apart were dropped from the follow-up analysis. Therefore, all participants with at least 2 valid QFT collected at least one month apart were included in the followup analysis. Fisher's exact test and logistic regression were used to assess associations of study sites and age, time between tests, or job position, respectively with conversion or reversion of follow-up tests. Signi cant associations were those that had a p-value <0.05 for all analysis.
Ethical clearance Permission to conduct the study was obtained from the Ethical Committee of the Kyrgyz Ministry of Health (registration no 01-3/270 from Sep 11, 2017 were included. There were ve HCW that did not provide age and were not included in the age-controlled analysis. There was no random effect of site on the outcome of the QTF test (p=0.063) based on the mixed effects logistic regression including age and job position as dependent variables for QFT when analysing all four sites independently. However, there were differences in the mixed effects logistic regression when comparing the NTC and the three other sites combined (p=0.02). Additionally, the average age of HCW at the NTC (44.7 years) was lower than the other sites combined (48.9 years; p<0.001); therefore, logistical analyses were split by site (Table 1). In univariate analysis, increasing age per year was associated with a positive quantiferon result (OR=1.02 per year, 95% CI=1.00-1.03, p-value<0.001), and this effect occurred among all job positions. Age was controlled for in the analysis due to the association between job position (p<0.001) and QFT positivity.
There were 189 (46.7%) HCW that had a positive QTF result. At both the NTC and the other sites combined, nurses had the highest level of recruitment to the study and ranged in age from 21-70 years old (Table 1). In the NTC, physicians (OR=8.7, 95%, CI=1.2-60.5, p=0.03) and laboratory staff (OR=19.8, 95% CI=2.9-135.4, p<0.01) had increased odds to have positive QFTs than administrative staff. Nurses (OR=5.9, 95% CI=1.0-35.4, p=0.05) and cleaners (OR=6.01, 95% CI=0.9-39.4, p=0.06) showed trends towards increased odds of positivity (Table 1). In the conservative pairwise comparisons for all job positions with the Bonferroni correction, no position was any different than another for testing positive. However; without the adjustment additional differences between positions were that lab workers had higher odds of being positive than cleaners (coe cient difference = 1.19, unadjusted 95% CI=0.09-2.29). There was no effect of job position on the QTF results for Karabalta, Karakol and Kemin combined (Table 1).  Table 2 Logistic regression results for differences between departments within the National Tuberculosis Centre (NTC), when controlling for age. had no effects on the follow-up status. There was no association with having a borderline positive rst test and the outcome of the 2 nd test being either positive, or negative (p=0.16). Of the 30 participants that had a 3 rd valid QFT performed, 9 of 10 (90%) and all (13/13, 100%) participants of the C and P status, respectively, remained positive. Of those with N status (6/6, 100%) remained negative for the 3 rd QFT (Table 3). Table 3 Follow-up QuantiFERON-TB Gold plus tests of HCW indicating stable and instable test results for HCW. Stable status was assigned when the 3 rd test matched the 2 nd , and instable status when it did not.

Discussion
Our study is the rst of its kind to describe prevalence and risk factors of LTBI among HCW in Central Asia. It supports the commonly reported ndings that job position in the hospital has an effect on QTF positivity, and there is a higher prevalence of LTBI among HCW than the general public (11,13,14). The prevalence of 46.7% QFT positive HCW in the current study was markedly higher than the LTBI prevalence of 3.9% reported from low-incidence countries (10) and 12% to 28% reported from the medium-incidence country Iraq (7), but well within the range of the 34 -60% LTBI among HCW reported from other highincidence countries (11). It should be considered that the cited studies investigated HCW in multi-disciplinary hospitals, while the participants of the current study were exclusively working with TB patients.
WHO guidelines for infection control among HCW were released in 2005, which have since been updated in 2019 (15), and have also been implemented in most health care institutions in Kyrgyzstan which should have had a meaningful impact on transmission and prevalence of LTBI (9,16). However, the current study indicates HCW have an increased risk of LTBI, especially doctors, laboratory personnel, those working in the drug resistance and smear-negative ward of the major tuberculosis hospital (NTC) in the Kyrgyz Republic; although, these differences are no longer signi cant when adjusting for multiple comparisons in pairwise analysis. Research conducted in Brazil, South Africa, and similar high TB burden countries have found that there is not a lack of protocols or equipment leading to gaps in infection control, but rather a lack of understanding of and adherence to infection control standards, uptake of prevention training, and e cient monitoring (17)(18)(19). HCW of the DR-TB and smear negative pulmonary TB departments had signi cantly higher odds of being QFT positive, although infection control measures should be particularly strict in the DR-TB ward, and risk of infection relatively low in the department of smear negative TB. Our ndings suggest that work safety conditions might not yet be optimal in Kyrgyz TB health facilities, but also that HCW are not yet fully compliantly following the infection control standards especially where the perceived risk of transmission is presumed to be comparably low.
Of the 49 HCW that provided valid follow-up QFT during the 17-month study period, 15 (30.6%) converted from negative to positive which may re ect the high risk of infection among these professions. These results correspond to ndings in previous studies whereby 11% of HCW in a medical institute in a rural area of Central Indian (incidence rate of over 300 / 100'000) converted with QFT over a period of one year (20) and 10% of HCW converted in Malaysia with an incidence rate around 63 per 100'000 population (21). Although serial testing with IGRA can be in uenced by various factors (20,22), QFT has repeatedly demonstrated its superior reliability over TST in systematically BCG vaccinated populations with high incidence of TB (23). Although there was no association with age, site, time between tests, or job position, the majority of patients that converted to positive remained positive at follow-up testing, strongly indicating new LTBI infections which may progress to active TB. In a setting with one third of TB patients having multi-drug resistant forms of the disease, active TB among HCW and their potential to spread the infection to vulnerable patient groups is of particular public health concern.
There were ve individuals with tests that reverted from positive to negative at follow-up, and it has been suggested that borderline results (falling within an "uncertain range" of e.g. 0.22-0.71 IU/mL) may be more likely to change at follow-up (20,24); however, we did not see an association with borderline positivity of the rst and the outcome of the second test. The apparent randomness and inconsistency of reversions at follow-up has been identi ed in previous research (25,26) and changes have not been explained by occupational exposure in HCW (27). Therefore, reversions identi ed in serial QTF testing should be interpreted with caution.
The ndings of this study are of particular impact for public health and TB control projects in Central Asia as they provide the rst insight in occupation speci c prevalence of LTBI; however, certain limitations of the study should be kept in mind when interpreting the results. Firstly, there is a lack of information regarding the duration that the HCWs have been working in health, particularly the TB sector. It has been documented in previous research that there is an increased risk of positivity as length of employment in pulmonary hospitals increases (14). This risk factor should be considered in future studies investigating TB transmission to HCW in Central Asia and the effectiveness of implemented control measures. It should be considered that the follow-up test interval of at least one month in the current study is much shorter than most studies where investigations for serial QTF testing were conducted at intervals of three (24) to 18 months (20). However, this effect on the results should be minimal as time between tests was not found to predict conversion status.
Finally, there is likely a high amount of variability among HCW at the NTC, as the wide con dence intervals are not re ected in the smaller sample size of the three other hospitals combined. Additionally, apparent differences between job positions for risk of QFT are no longer signi cant when adjusting for multiple comparison, and further research is required to fully understand HCW risk.
Our study had several limitations. Due to its relatively small sample size, 95% con dence intervals were large for some groups of HCW weakening statements regarding the risk of LTBI although ORs might be well above 4.0 and trends might be observable. Larger sample sizes should be considered in future studies when investigating health care facilities in Central Asia. As the baseline positivity rates of urban and rural populations are unknown for Kyrgyzstan, we had to use administrative staff as base group which might however bring in some bias. Our study has focussed on TB hospitals and laboratories only; population based studies would help estimate the real LTBI risk of HCW compared to the normal population in similar environment. Studies in other types of hospitals and health facilities would increase the representativeness of the results for all Kyrgyz HCWs. Simultaneous