Risk of tuberculosis and uptake rates of latent tuberculosis infection screening among clinical risk groups in South Korea: A nationwide population-based cohort study

Background This study aimed to investigate actual tuberculosis (TB) risk and uptake rates of latent tuberculosis infection (LTBI) screening among eight clinical risk groups specified in Korean guidelines. Proportions of potentially preventable TB in these groups were also calculated. Methods and Findings Patients enrolled before January 1st, 2018, were classified into a prevalence cohort whereas those enrolled thereafter were classified into an incidence cohort. Both cohorts were followed up until December 31st, 2020. Sex, age, and calendar year-adjusted standardized incidence ratio (SIR) of tuberculosis was calculated with total population in South Korea as a reference group. The number of TB patients notified in 2018 was investigated for both prevalence and incidence cohorts. SIR of TB in each incidence cohort was higher than that in each corresponding prevalence cohort. Among all incidence cohorts, SIR in people living with human immunodeficiency virus (PLHIV) was the highest (17.41, (95% CI: 14.14-21.43)). Although classified as moderate TB risk diseases in current guideline, end-stage renal disease (ESRD) (8.05, (7.02-9.23)) and uncontrolled diabetes mellitus (DM) (6.31, (5.78-6.99)) showed high SIRs comparable to other high-risk diseases. Among total TB cases notified in 2018, each cohort accounted for less than 1.5% except for patients with DM. The uptake rate of LTBI test was the highest among patients using TNF inhibitors (92.7%), followed by those who underwent organ transplantation (60.4%) and PLHIV (41.3%). Conclusions LTBI screening should be reinforced for certain clinical risk groups such as ESRD or uncontrolled DM. Beyond the current guideline, additional high-risk groups should be identified.


INTRODUCTION
Approximately a quarter of population in the world is presumed to be infected with 51 tuberculosis (TB) [1]. For TB elimination, strategies to tackle the large TB reservoir are essential [2]. 52 However, current diagnostic tools for latent tuberculosis infection (LTBI) have low predictive 53 values [3], which can result in high numbers needed treat to prevent a TB case. Several conditions 54 raising TB risk have been investigated [4]. Previous guidelines on LTBI have commonly specified two 55 key groups: TB contacts and clinical risk groups [5][6][7][8]. 56 TB incidence in South Korea has continuously decreased since 2000. TB incidence was 96.3 57 cases per 100,000 population in 2000 and 44.6 cases per 100,000 population in 2021 [9]. With a 58 decrease in TB incidence, strategy to prevent reactivation has been underscored as in other low-59 incidence countries [10,11]. However, in contrast to contact investigation which is managed by the 60 government [12], LTBI screening in clinical risk group is usually performed in private sector, which 61 accounts for more than 90% healthcare facilities in South Korea[13]. Although clinical risk groups 62 have been specified in Korean guidelines since amendment in 2014 [14], actual TB risk and uptake 63 rate of LTBI screening in these groups have not been evaluated before. 64 Thus, the objective of this study was to investigate the actual TB risk among high-risk 65 groups specified in Korean guidelines for TB and uptake rate of LTBI screening in each group. 66 Additionally, proportions of potentially preventable TB by implementing LTBI screening in these risk 67 groups were calculated among total nationwide TB burden.  Table). Each group was denoted as follows: Group 1, 74 people living with human immunodeficiency virus (PLHIV); Group 2, patients who underwent solid 75 organ or hematopoietic stem cell transplantation; Group 3, patients who used tumor necrosis factor . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023. Patients enrolled in each cohort were followed up from January 1 st , 2018 for the prevalence 92 cohort and from the date of enrollment for the incidence cohort. Follow-up was terminated at 1) date 93 of TB diagnosis, 2) date of death or 3) 31st December 2020, whichever came first. Incidence rate of 94 TB and mortality were calculated for each cohort. To estimate TB risk, sex, age, and calendar year-95 adjusted standardized incidence ratio (SIR) of tuberculosis were calculated with total population in 96 South Korea as a reference group. TB incidence among total population was calculated using database 97 of KNTSS notified from 2018 to 2020. SIR was presented after stratifying subjects into three age  Age-stratified proportion of each prevalence and incidence cohort among nationwide TB 101 patients notified in 2018 was calculated. For the incidence cohort, only patients enrolled in 2018 were 102 analyzed.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023

103
The percentage of those who underwent interferon-gamma release assay (IGRA) or 104 tuberculin skin test (TST) was investigated. Screening was classified by the date of LTBI 105 examination. Recent screening denoted that the screening was performed within one year before the 106 date of enrollment. New screening indicated that the screening was performed at the date of 107 enrollment or thereafter. Past screening represented that the screening was performed more than one 108 year before the date of enrollment. As some patients such as those who were scheduled for organ 109 transplantation and those who were going to use TNF inhibitors were requested to undergo LTBI 110 screening in advance, the percentage of recent or new screening was presented as one of the major 111 outcomes of this analysis.

RESULTS
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Table. 131 Among eight prevalence cohorts, Group 4 showed the highest TB incidence (487.4 per 100,000 132 person-years) and mortality rate (9592.9 per 100,000 person-year). Similarly, among incidence 133 cohorts, TB incidence (1117.4 per 100,000 person-years) and mortality (19619.5 per 100,000 person-134 years) were the highest in Group 4 (Table 1). In every group, TB incidence was higher in the 135 incidence cohort than in the prevalence cohort. Mortality was higher in the incidence cohort than in 136 the prevalence cohort for all groups except for Group 8.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023. in elderly population were lower than those in younger population.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

165
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023. Among all groups, SIR of TB was higher in the incidence cohort than in the prevalence 167 cohort (Table 3). Group 1 had the highest SIR (17.41, 14.14-21.43), followed by Group 3 (9.67, 7.71-168 12.12), Group 2 (8.90, 7.35-10.77), and Group 4 (8.05, 7.02-9.23), which comprised the high-TB risk 169 group. Among other groups, Group 8-1 and Group 7 showed considerably high SIRs (6.31, 5.78-6.99 170 and 6.07, 5.46-6.75, respectively). As in the prevalence cohort, SIRs in elderly population were lower 178 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

187
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023.  accounted for less than 1.5% of total notified TB cases except for Group 8 (17.2%). The proportion of 192 each incidence cohort was similar to or lower than that of the prevalence cohort. Most of incidence 193 cohorts accounted for less than 1% of total notified TB cases except for Group 8 (3.3%).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023. Age-stratified proportion of each cohort was presented in Table 4. For high TB-risk diseases 205 (Group 1, 2, 3) and DM (Group 8) proportions of incidence cohort were highest among TB patients 206 aged between 35-64 years. For moderate TB-risk diseases such as ESRD (Group 4), or malignancy 207 (Group 5, 6, 7), those were highest among TB patients aged over 65 years. Proportions of each 208 incidence cohort were relatively low among TB patients aged under 35 years, in most of diseases.

209
However, those of high TB-risk disease (Group 1, 2, 3)  groups were even lower. IGRA rather than TST was the mostly used LTBI test in each group.

219
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

228
Recent screening denotes that latent TB examination (IGRA or TST) is performed within one year before the 229 date of enrollment. New screening indicates that the screening is performed at the date of enrollment or 230 thereafter. Past screening represents that the screening is performed more than one year before the date of 231 enrollment. 232 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

234
In this study, well-known high TB risk diseases showed high SIRs as expected. Patients with 235 ESRD showed TB risk comparable to those with high TB risk diseases. TB risk was relatively low in 236 patients who underwent gastrectomy and patients with DM among groups with moderate TB risk.

237
However, patients with uncontrolled DM status showed a relatively high TB risk. The uptake rate of 238 LTBI screening in these clinical risk group was still suboptimal in all groups except for patients who 239 used TNF inhibitors.

240
Previous LTBI guidelines did not specify when to treat in detail [5][6][7][8]15]. We demonstrated 241 that TB risk of patients who developed the disease newly was higher than that of those who were 242 diagnosed several years ago. This might be attributable to a decreased immunity of patients with an  Among patients with malignancy, it was presumed that immunosuppressive states might be temporary 251 during a session of anticancer treatments such as chemotherapy just after the diagnosis of cancer, 252 which was demonstrated by decreasing SIR of TB with increasing time after cancer diagnosis [19].

253
Our findings underscored that LTBI screening should be focused on incident groups.

254
However, most of incident groups showed higher mortality as well as higher TB incidence than 255 prevalent groups. Among patients who were newly diagnosed as ESRD, 33.0% (3617/10977) of 256 patients died during the follow-up period. Considering adverse effects of LTBI treatment such as 257 gastrointestinal trouble and hepatotoxicity[20], we speculate that providing LTBI treatment to these 258 critically ill patients might be unfeasible in many cases. Moreover, in patients newly diagnosed as 259 malignancy, preventing TB would not be a medical priority. In a previous study, TB incidence was . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 260 higher among patients with malignancy such as esophageal, lung, pancreatic cancer, and multiple 261 myeloma, showing a relatively low 5-year survival [21]. We expect that the acceptance rate for LTBI 262 treatment among these patients with limited life expectancy would be low. Low LTBI screening 263 uptake rate in patients with malignancy (Groups 5, 6, 7) might be related to this reason.

264
However, Groups 2 and 3 showed relatively high uptake rates for LTBI screening (60.4% 265 and 92.7%, respectively). This might be because LTBI screening and treatment are more feasible in 266 these groups than in other groups. Mortality rate of Group 3 was the lowest among all incident groups.

267
This finding suggests that LTBI screening would be more feasible when they are scheduled to be a 268 high-risk group, than when they had already become. Relatively low uptake rate of LTBI screening 269 rate among PLHIV (41.3%) who were specified as the highest TB risk group could be explained by 270 this hypothesis considering a low diagnosis rate of HIV and a low CD4+ cell count at diagnosis 271 suggesting delayed diagnosis of HIV in South Korea[16,17]. We presume that diagnosis of HIV in 272 earlier course of the disease would enhance the LTBI screening uptake rate. Further studies are 273 needed to verify this hypothesis.

274
ESRD showed high SIR as other diseases with high TB risk, although it was specified as a 275 disease with moderate TB risk in South Korea[15]. Additionally, it showed the highest SIR among all 276 prevalent groups, suggesting that immunosuppression state could last for a longer period than other 277 groups. Therefore, it should be reclassified as a disease with a high TB risk. However, LTBI 278 screening is not widely performed for those with ESRD. We speculate that screening LTBI at earlier 279 stage in chronic kidney disease (CKD) could be an alternative option for increasing the uptake rate of 280 LTBI screening, like that for PLHIV. However, potential nephrotoxicity of anti-TB medication, 281 especially rifampicin, is a concern[22], which might lower the uptake rate of LTBI treatment among 282 patients with pre-dialysis CKD. LTBI regimen without potential risk of nephrotoxicity should be 283 investigated.

284
DM is a known risk factor for TB [23]. However, in our study, TB risk was relatively low 285 among diseases with moderate TB risk. Instead, uncontrolled DM status rather than disease itself 286 contributed to TB development, as reported in a previous study [24]. By focusing on patients with . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 5, 2023. . Second, when compared to SIR in younger generation, SIR in elderly population was 300 much lower in most groups. This suggests that identifying diseases with a high TB risk could be a 301 useful tool for risk stratification among younger generation whereas it is less useful in elderly 302 population. The low SIR in elderly population is attributable to a high TB incidence among general 303 elderly population in South Korea[27]. Other comorbidities not specified in current guidelines, 304 waning immunity derived from aging, and malnutrition might also contribute to this [28]. In addition, 305 we demonstrated that proportions of high TB risk diseases among elderly TB patients were extremely 306 low, and lower than other age groups, implying LTBI screening strategy targeting for high TB risk 307 diseases is less efficient in elderly population, than in other age groups.

308
In the previous Canadian study, covering immigrants from high TB burden countries 309 potentially prevented 37.1% of total TB cases [25]. However, in South Korea, the proportion of 310 foreign-born residents among total population was 3.4% in 2020, which was lower than average 311 (14.7%) of other high-income countries [29]. Instead, native elderly TB patients are a key group in 312 South Korea. In 2021, TB patients who aged 65 years or above accounted for approximately 51.0% of 313 total TB cases [9]. LTBI screening and treatment among elderly population are not routinely . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2023. ; https://doi.org/10.1101/2023.06.02.23290863 doi: medRxiv preprint 314 recommended due to low predictive values of diagnostic tools such as IGRA and TST and higher 315 incidence of adverse effects during LTBI treatment [20]. Recently, the necessity of expanding LTBI 316 screening to elderly population has been suggested[30] and the feasibility of LTBI treatment among 317 high-risk elderly population has been reported [31]. Further studies identifying high-risk elderly 318 population should be implemented.

319
This study has a strength of linking two databases covering the entire South Korean 320 population. Thus, many study subjects were included. We compared several diseases with high or 321 moderate TB risk simultaneously at nationwide level within the same study period. However, our 322 study had a limitation in that the LTBI status of each patient was unavailable. SIR among patients 323 with LTBI might reflect more accurate TB risk. However, considering that most diseases did not 324 affect LTBI prevalence [5] and that age was the most significant factor associated with LTBI 325 prevalence[32], we assumed that LTBI prevalence in each group and general population were not 326 quite different, and that age-adjusted SIR would be sufficient for estimating the actual TB risk.

327
Another limitation was that the number of patients with LTBI was not known. Thus, calculation of 328 further cascade of care such as initiation of LTBI treatment was unfeasible.

329
In conclusion, LTBI screening in certain clinical risk groups such as patients with ESRD and 330 patients with uncontrolled DM should be reinforced. Ideally, LTBI screening should be provided 331 around the date of the disease diagnosis. However, feasibility of LTBI treatment at that period 332 remains a problem. Beyond the current guideline, identification of additional high-risk groups, 333 especially among elderly population, is required. Prevention Agency (https://www.kdca.go.kr/) (2020E310100 to JSK). The funders had no role in 338 study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2023. ; https://doi.org/10.1101/2023.06.02.23290863 doi: medRxiv preprint