Trend and treatment outcomes of latent tuberculosis infection among migrant persons in Japan: analysis of Japan Tuberculosis Surveillance data

Screening for latent tuberculosis infection (LTBI) among migrant population has become a critical issue for many low TB burden countries. Evidence regarding effectiveness of LTBI programs are limited, however, partly because of paucity of national data on treatment outcomes for LTBI. In Japan, notication of LTBI is mandatory, and its treatment outcome is reported as part of Japan’s national TB surveillance system. We thus conducted a detailed analysis of LTBI among foreign-born persons, to update the epidemiological trend of newly notied LTBI between 2007 and 2018, and, to examine the treatment regimen and outcome of those notied in 2016 and 2017, focusing specically on the potential risk factors for lost to follow-up.

Foreign-born populations contribute considerably to both the number and proportion of all tuberculosis (TB) cases, especially in countries with low TB incidence [1][2][3]. In Japan as well, which is a TB middleburden country with a noti cation rate of 12.3 per 100,000 in 2018, although the proportion of foreignborn persons out of all TB cases was 9.5%, that among those aged between 15 and 24 years old has reached 70.8% in the same year [4]. It has been pointed out that the majority of these TB cases occurs due to the reactivation of latent tuberculosis infection (LTBI) acquired in their country of origin. As a large proportion of migrants is born in high-TB burden countries, where it has been estimated that approximately 26-46% of the population are latently infected with TB [5,6], screening for not only active TB but also LTBI among the migrants, has become a critical issue for many low TB burden countries (< 10 cases/100,000 population). In Japan as well, 58.2% of foreign-born TB patients come from 5 countries in Southeast Asia, with a TB incidence of more than 100 per 100,000 population.
Yet in its recently revised guideline, the World Health Organization (WHO) has only conditionally recommended LTBI screening among migrants living in low TB burden countries [7]. The guideline points to the low-quality evidence for the effectiveness and cost-effectiveness of LTBI programs in these settings, including challenges in achieving treatment adherence and completion rate. Studies reporting national data on treatment outcomes for LTBI are however limited, due mainly to LTBI not requiring mandatory noti cation in many countries. Japan is however one of the few countries where noti cation of LTBI is mandatory under the Infectious Diseases Control Law, and its treatment outcome is evaluated and reported as part of Japan's national TB surveillance system (JTBS). The authors have previously estimated the treatment outcomes of LTBI noti ed between 2007 and 2014 -the outcomes were however determined using proxy variables because the JTBS then only evaluated treatment outcome for pulmonary TB [8]. In 2017, the JTBS underwent a system revision, making cohort analysis possible for all types of TB, including extra-pulmonary TB and LTBI. We thus conducted a detailed analysis of LTBI among foreign-born persons, with two main objectives -rstly to update the epidemiological trend of newly noti ed LTBI between 2007 and 2018, and secondly, to examine the treatment regimen and outcome of those noti ed in 2016 and 2017, focusing speci cally on the potential risk factors for lost to follow-up among foreign-born LTBI patients.

Methods
Japan introduced its rst electronic surveillance system in 1987, which since then has undergone several major system revisions. Details of the system can be found elsewhere [4]. Brie y, at the end of each year, it produces four sets of data; 1) a list of all active TB and LTBI cases newly noti ed in that year ("newly noti ed dataset"), 2) a list of all active TB and LTBI cases registered at the end of the year ("end of year dataset"), 3) a list of active TB and LTBI cases noti ed in the previous year, with treatment outcomes ("cohort dataset"), and 4) a list of all cases who were de-registered in that year ("de-registered dataset").
We thus analyzed the "newly noti ed dataset" to examine the overall trend of noti cation and by age groups and modes of detection between 2007 and 2018, and the "cohort dataset" for treatment regimen and outcomes of foreign-born persons noti ed with LTBI in 2016 and 2017. Potential reasons for lost to follow-up were further explored, and logistic regression analysis was conducted to identify risk factors.
Comparisons were made with the Japan-born patients where appropriate. The de nitions for the main variables can be found in the Annex (S1 Appendix). R version 3.6.3 (R Development Core Team, Vienna, Austria) was used for all statistical analyses.

Results
Trend in LTBI noti cation Figure 1 shows the annual trend of the number of overall LTBI noti cations by country of birth, and the proportion of foreign-born persons among the total. The number of total LTBI noti cations has reached a peak in 2011, and since then it has declined and in the past ve years or so, has stabilized around 7,000 a year. However, both the number and the proportion of foreign-born LTBI patients have increased steadily, reaching 963 cases in 2018 (13.0%, 963/7,414). Looking at the trend in the noti cation by age groups, while among the Japan-born patients, the number of noti cations has increased only among those aged 55 years old and above, among the foreign-born patients, the number has increased rapidly among all age groups except those aged between 0 and 14 years old (Fig. 2). Regarding the modes of detection, for both Japan and foreign-born patients, the majority were detected via contact investigation over the study period (foreign-born; 80.3%, and Japan-born; 72.3%). However, while among the Japan-born, only the number of those detected at hospital has increased, among the foreign-born, the numbers of those detected via contact investigation, routine screening, and at hospital all similarly expressed steady increase (Fig. 3).

LTBI treatment regimen
In the "newly noti ed dataset", between 2016 and 2017, a total of 12,743 Japan-born and 1,407 foreignborn persons were noti ed with LTBI. From the "cohort dataset", treatment regimen and outcome were available for 12,817 Japan-born and 1,418 foreign-born LTBI patients (the numbers from the cohort dataset do not necessarily match with those of the newly dataset, due to various reasons such as delay in entering patient data into the system). Of the 12,817 Japan-born and 1,418 foreign-born LTBI patients, 96.0% (n = 12,298) and 92.3% (n = 1,309) had started their treatment with INH monotherapy, respectively.
Higher proportion of foreign-born than Japan-born patients had started their treatment with RFP monotherapy (5.5% vs. 1.8%, p < 0.001) ( Table 1). Treatment outcomes of those who have started treatment, i.e. excluding those whose treatment regimen was recorded as "no treatment", are summarized in Table 2. The proportion of lost to follow-up (8.2% vs. 8.4%, p = 0.78) were similar among the Japan-and foreign-born patients, however, proportion of transferred-out was much higher among the foreign-than Japan-born patients (8.3% vs. 1.5%, p < 0.001), making the overall treatment success slightly higher among the Japan-born patients (85.8% vs. 82.2%, p < 0.001). However, of patients who have transferred-out, while among the Japan-born, 94.8% (183/193) were transfer-out within Japan, among the foreign-born, 43.0% (50/116) had transferred out of Japan.
Considering that treatment outcome of those who have been transferred outside of Japan is practically non-traceable, these were re-categorized as lost to follow-up -this had the effect of making lost to followup considerably higher among foreign-than Japan-born patients (12.0% vs, 8.2%, p < 0.001). Proportions of lost to follow-up and those who have transferred out of Japan (n = 167), among various socio-demographic characteristics within foreign-born persons are summarized in Table 3. The results of logistic regression analysis, whereby all of the above characteristics were entered as explanatory variables, are summarized in Table 4.

Discussion
Our results indicated that LTBI noti cations among the foreign-born persons have continued to increase, although this is not surprising considering that the number of foreign-born persons with active TB cases is also on the increase [4]. Among the Japan-born patients, the rise was mainly seen among the older persons, which can be attributable largely due to the abolition of age limit for LTBI treatment in the national guideline in 2010 [9] and an increasing number of elderly persons being tested for and diagnosed with LTBI before being treated for other medical conditions such as rheumatoid arthritis (i.e. hence LTBI being detected "at hospital"). On the other hand, among the foreign-born persons, the LTBI noti cations increased in all age groups except those aged between 0 and 14 years old, and among all modes of detection. Among the Japan-born persons, "routine" screening for LTBI has only really been conducted at selected workplaces, such as among high-risk healthcare workers -however, the constant increase in the noti cation of LTBI among foreign-born persons via "routine screening" may indicate that increasingly, schools and workplaces are making individual decisions to introduce LTBI screening as part of the routine health-check for foreign-born persons. Indeed, there have been sporadic reports of LTBI screening being conducted as part of medical examination upon admissions to universities [10,11]. In the absence of a clear national guideline on priorities for LTBI among foreign-born persons in Japan, rigorous studies on effectiveness and cost-effectiveness are urgently needed to determine the most appropriate intervention for case nding of LTBI targeting foreign-born persons in Japan.
As for the issue of treatment adherence, to our knowledge, this is the rst detailed analysis of LTBI treatment outcome among foreign-born persons at a national level in Japan. Outside of Japan, several systematic reviews on adherence for and outcomes of LTBI treatment have been conducted -for example, a review that was conducted for the WHO 2015 LTBI Guideline has concluded that treatment completion rates varied across different risk groups, ranging from 6-94%, with lower completion rates for prisoners and immigrant [12]. Another study which looked at studies from US and Canada has reported completion rates to be ranging between 22% and 90% [13], and a more recent review reported rates between 7-86%, both for foreign-born person [14]. A meta-analysis that was published in 2016 has estimated the pooled treatment completion rates for migrants to be 14.3% [15]. The studies which were included in these systematic reviews do, however, vary considerably in terms of sample size, study design, types of immigrant and treatment regimen, and hence caution is required when interpreting the results.
Our study results indicate the treatment completion rate for foreign-born persons in Japan is relatively high, compared to other studies included in the abovementioned reviews, albeit not reaching the national target of 85%.
Many studies have also examined the predictors for adherence to LTBI treatment, and in general, have tended to conclude that demographic factors such as age, sex, place of birth and race do not seem to in uence completion rate [13]. In our study too, neither sex nor age were conclusive risk factors for lost to follow-up among foreign-born LTBI patients. However, as for age, although not statistically signi cant, there was a clear tendency for the risk to increase with age. This is quite understandable, as younger children are usually overseen by their guardians and are therefore less likely to become lost to follow-up. Other socio-economic factors, such as unemployment and lack of health insurance, have previously been associated with failure to complete [16,17] -in our study, none of the foreign-born LTBI patients who had started treatment were reported as "non-insured". Receiving social welfare assistance and having "others and unknown" for insurance status, which could include short-term visitors with private health insurance, were associated with elevated risk but were not statistically signi cant. As for job status, being employed on a daily or temporal basis, and having "others and unknown" for job status, were both associated with being lost to follow-up. Considering that LTBI treatment is publicly subsidized and is unlikely to place a signi cant nancial burden for the patients, being employed on a daily or temporal basis may represent not necessarily poor economic condition but more an unstable lifestyle and greater mobility which are putting patients at risk of becoming lost to follow-up.
As for the length of treatment regimen, while the WHO 2015 Guideline has concluded that longer treatment duration was detrimental to treatment completion, a systematic review that was published a year later has concluded that study results were inconclusive, with some showing better treatment outcomes for shorter regimens using rifampicin, pyrazinamide, rifabutin and/or INH, than the standard regimen (6 or 9 months of INH) while others showing similar completion rates, and also that the studies themselves were too heterogeneous to conduct pooled analysis [14]. In our study, the proportion of lost to follow-up was bigger among those who have started treatment with INH than those with RFP monotherapy (12.3% vs. 6.4%). The difference was shown not to be signi cant in the logistic regression analysis (p = 0.12), however, this could possibly be due to small number of those on RFP monotherapy. It must also be noted that the information regarding treatment regimen is only of that upon noti cation -in reality, regimen can and do change during the course of the treatment, however, neither the change nor the new regimen after change is captured in the JTBS.
It has been reported that one of the major challenges in ensuring adherence especially for LTBI treatment is to overcome the psychological resistance held by patients to take drugs for a non-contagious and nonsymptomatic infection that may never develop into active disease, but which could cause potential adverse effects, and convince them of the potential bene ts of prevention [13]. This is could be especially challenging especially when targeting patients who come from culture that is unfamiliar with the concept of screening and prevention [18]. For example, a prospective study that has examined the predictors for non-completion of LTBI treatment has concluded that a perceived risk of progression to active TB is strongly associated with better adherence [19]. However, discussions regarding which factors may in uence the construction of such perceived risk and bene t are still inconclusive. One possible factor is the status of being a contact of an active TB patient, and having a close experience with the disease. Indeed, some studies have indicated that contacts tended to have higher completion rates than other population groups [20,21,22]. In our study, whether or not the patient was a contact of a TB case could be identi ed from the variable "mode of detection" -those who were detected by contact investigation are obviously case contacts, however, our results indicated that being a contact was not associated with better treatment adherence. This may suggest that being a contact of a TB case alone is a weak motivating factor to adhere to and complete LTBI treatment, at least among foreign-born persons in Japan.
Another potential factor is patient education and counselling -for example, studies from prison have indicated that those who have received education sessions about LTBI prior to being released had higher treatment completion rates post-release, than those who did not [23]. Several studies have shown that well-designed educational intervention using culturally and socially sensitive languages can assist patients make better-informed decisions about the potential risks and bene ts of LTBI treatment [24,25].
We were unable to assess the possible impact of education as such is not collected in the JTBS. However, numerous studies on adherence support to foreign-born active TB patients in Japan have pointed to the challenge of language [26,27]. Even today, Japan continues to be a largely mono-cultural society and it is quite common that a large majority of local healthcare workers only speak Japanese. There is an increasing awareness of the need to improve the language capacity of health services, against the rising number of foreign-born patients in Japan -however, the resources are still very much limited. Under such situation, it is not di cult to imagine that foreign-born patients in Japan are not receiving the appropriate and adequate information that they need to make informed decisions about LTBI and treatment. Studies are needed to explore the most effective information, education, and communication for foreign-born patients regarding LTBI.
Finally, we considered the time between entry to Japan and diagnosis of LTBI as a potential risk factor for lost to follow-up. Compared with those who had been diagnosed within 2 years of arriving to Japan, those who had been in Japan for more than 2 years but less than 5 years had a slightly raised risk of becoming lost to follow-up (adjusted odds ratio 1.69, 95% con dence interval 1.00-2.87). The risk probably re ects the risk of transferring-out of Japan, rather than becoming lost to follow-up in Japan, as the proportion of international transfer-out was the highest among those who had been in Japan for more than 2 years but less than 5 years (n = 15/222, 6.7%). The reasons are not clear, though this may be related to the duration of permit to stay, which majority of foreign-born students and workers are requested to apply when staying in Japan. The duration of stay depends on the type of permit, however, it is usually 1 year, 3 years or 5 years. In other words, those who are transferring out of Japan may simply be doing so, because their permit to stay is expiring. This brings up another important issue surrounding the continuum of care for LTBI patients across national borders, but the discussions are beyond the scope of this study.
Our study is not without limitations. Firstly, as our sole source of data was the JTBS, we were unable to examine the effects of variables which were not collected in the JTBS, such as frequency of side-effect [28,29], alcohol and/or drug dependence [26,30], and prior BCG vaccination (data regarding prior BCG vaccination is only collected for infants) [31], which have previously been identi ed as being potentially detrimental to treatment adherence. Secondly, our study did not take into account those who have become lost to follow-up at various stages in the cascade of care in diagnosis and treatment of LTBI, including initial testing of those intended for screening, completing medical evaluation if test was positive, and initiating treatment. For example, the previously mentioned meta-analysis has concluded that screening completion rates was the lowest among the migrants (43%), compared with other populations such as medical personnel (86.1%), marginalized persons (83.3%) and contacts of active case (79.3%) [15]. A separate study is probably required to examine the entire cascade of care for LTBI among foreign-born persons in Japan.

Conclusions
Migrant population continues to have a considerable impact on TB epidemiology globally, and in Japan as well, and thus it is vital that active measures are taken not only to control TB but also LTBI. While the WHO updated guideline for LTBI management states that "concerns about adherence should not be a barrier to use of preventive treatment" [7], the authors nevertheless believe that clinical bene t to individual patient and the success of control programs are subject to completion of LTBI treatment, and hence interventions to improve adherence are a critical component of programmatic management of LTBI.
As previous studies have already shown, efforts to establish a "one-size ts-all" approach to improve adherence to LTBI treatment are probably not likely to succeed. Our results have indicated risk factors for lost to follow-up, some of which were potentially unique to Japan. On the other hand, we have also pointed to the need for a further, in-depth study to explore the socio-economic and cultural situation in Trend of LTBI noti cation by mode of detection, 2007-2018