Prevalence of and risk factors associated with latent tuberculosis in Singapore: A cross-sectional survey

Accepted Manuscript Title: Prevalence of and risk factors associated with latent tuberculosis in Singapore: a cross-sectional survey Authors: Peiling Yap, Kristin Hui Xian Tan, Wei Yen Lim, Timothy Barkham, Linda Wei Lin Tan, Mark Chen I-Cheng, Yee Tang Wang, Cynthia Bin Eng Chee PII: DOI: Reference:

S1201-9712(18)34412-6 https://doi.org/10.1016/j.ijid.2018.05.004 IJID 3234

To appear in:

International Journal of Infectious Diseases

Received date: Revised date: Accepted date:

8-3-2018 3-5-2018 4-5-2018

Please cite this article as: Yap Peiling, Tan Kristin Hui Xian, Lim Wei Yen, Barkham Timothy, Tan Linda Wei Lin, I-Cheng Mark Chen, Wang Yee Tang, Chee Cynthia Bin Eng.Prevalence of and risk factors associated with latent tuberculosis in Singapore: a cross-sectional survey.International Journal of Infectious Diseases https://doi.org/10.1016/j.ijid.2018.05.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Prevalence of and risk factors associated with latent tuberculosis in Singapore: a cross-sectional survey Running head: Latent tuberculosis in Singapore

Tan2, Mark Chen I-Cheng1,2, Yee Tang Wang4, Cynthia Bin Eng Chee4

SC R

Affiliations:

IP T

Peiling Yap1, Kristin Hui Xian Tan2, Wei Yen Lim2, Timothy Barkham3, Linda Wei Lin

1

Communicable Disease Centre, Institute of Infectious Diseases and Epidemiology, Singapore, Singapore 2

U

Saw Swee Hock School of Public Health, National University of Singapore, Singapore, Singapore Department of Laboratory Medicine, Tan Tock Seng Hospital, Singapore, Singapore

4

Tuberculosis Control Unit, Tan Tock Seng Hospital, Singapore, Singapore

M

A

N

3

ED

Corresponding author: Name: Chen I-Cheng Mark

PT

Full address: Tahir Foundation Building, 12 Science Drive 2, #09-01L, Singapore (117549)

CC E

Contact: [email protected]

Highlights

An overall latent tuberculosis prevalence of 12.7% amongst Singapore residents. A wide variation in positivity rate based on the residents’ country of birth. Risk factors include age, education and socio-economic status and alcohol usage

A

  

Abstract Objectives: We performed the first LTBI cross-sectional survey in Singapore, utilizing the QuantiFERON Gold In-tube (QFT-GIT) assay to collect data on the prevalence of LTBI and identify potential risk factors associated with it. 1

Methods: Nation-wide household addresses were randomly selected for enumeration, where Singaporeans or Permanent Residents (PRs), aged 18-79 years, were identified. One eligible member per household was selected using the Kish grid. Each participant answered a questionnaire, assessing medical history (including TB), socio-economic factors and lifestyle and provided a blood specimen for the QFT-GIT assay. Participants with positive QFT-GIT results were defined as having LTBI if they were asymptomatic. To identify independent risk factors, adjusted hazard ratios were obtained using the multivariable modified Breslow-Cox

IP T

Proportional Hazard Model.

Results: An overall QFT-GIT positivity rate of 12.7% was detected amongst 1682 Singapore

SC R

residents. There was a wide variation in the positivity rate according to the residents’ country of birth. Higher LTBI prevalence was also significantly associated with increasing age, lower educational and socio-economic status and alcohol use.

Conclusions: Given the high prevalence of LTBI amongst foreign-born residents from

U

regional countries, similar studies should be conducted amongst migrants in Singapore to

A

N

improve national guidelines on screening and preventive treatment against LTBI.

INTRODUCTION

ED

M

Keywords: latent tuberculosis, interferon-gamma release assay, prevalence, risk factors, population

Tuberculosis (TB) has killed more than a billion persons in the last 200 years and is currently

PT

the leading cause of death due to a single infectious agent [1]. The emergence of drug resistant strains and global migration have further complicated its control [2]. The World

CC E

Health Organization (WHO) estimated that there were 10.4 million new cases in 2016, with the WHO South-East Asia and Western Pacific Region accounting for 47% and 17% respectively of the global burden [1]. Latent tuberculosis infection (LTBI) is characterized by a state of continuous immune response to stimulation by M. tuberculosis antigens without

A

displaying clinical manifestations of active TB [3]. The lifetime risk of TB reactivation for an individual with LTBI is approximately 5–10% [3–5]. The latest global estimates suggest that about 1.7 billion people had LTBI in 2014 [6]. Targeted treatment of LTBI, in high risk groups, is the main intervention available to prevent development of active TB [1,7,8].

2

Singapore is an island city state of 719 square kilometres in South-East Asia. Its total population of 5.64 million comprises a resident population of 3.41 million citizens and 0.53 million permanent residents (PRs), and 1.7 million foreigners who are granted long-term passes to work, study or live as dependents in the country [9]. Following a decade of decline to a historical low of 35 per 100,000 population in 2007, Singapore’s TB incidence rate among its residents increased in 2008 and has remained at around 40 per 100,000 population [10,11] despite the efforts of the Singapore TB Elimination Programme (STEP) [10]. Possible

IP T

factors contributing to this stagnation include on-going community transmission due to delayed diagnosis of infectious cases, and reactivation of latent TB infection (LTBI) in a

SC R

rapidly aging population (who acquired their infection in the remote past when Singapore’s

TB rates were very high), or among newly inducted citizens or PRs, the majority of whom are foreign-born from surrounding high TB incidence countries [10,12]. Profiling the prevalence of LTBI amongst long-term residents in Singapore would allow us to estimate the potential

U

size of the current reservoir of infection and understand how this might affect the future

A

N

incidence of active TB [13].

In view of universal BCG vaccination at birth since the 1950s and a BCG re-vaccination

M

programme for school leavers from the 1950s to 2001, it is likely that a tuberculin skin test (TST) survey will overestimate the reservoir of LTBI in our local population [14,15]. In this

ED

study, we performed for the first time a large population-based survey of LTBI in Singapore, utilizing an interferon-gamma release assay (IGRA), the QuantiFERON Gold In-tube (QFT-

PT

GIT) (Qiagen, Hilden, Germany) assay. The study aimed to estimate the prevalence of LTBI stratified by key sociodemographic variables, and to compare these estimates with incidence rates of active TB. We also identified groups at risk for LTBI in the resident population of

CC E

Singapore to understand underlying drivers of TB epidemiology in Singapore.

A

MATERIALS AND METHODS Study design This was a cross-sectional study conducted between April 2014 and March 2015. The survey was nested within a larger population-based cohort study known as the Singapore Health Study 2 (SH2), which aimed to monitor the health of Singaporeans in preventive health and risk behaviours and chronic non-communicable diseases. 3

Study sites Across the island city state of Singapore [9], we set up 4 study sites: SATA CommHealth clinic at Jurong East, SATA CommHealth clinic at Woodlands, Cheng San Community Club at Ang Mo Kio and a data collection centre within Bras Basah Complex in the west, north, north-east and south-east central regions of Singapore respectively. All sites were similarly equipped with the necessary amenities for the study.

IP T

Study population

A random sample of 32,100 household addresses was selected from the National Database on

SC R

Dwellings in Singapore maintained by the Department of Statistics. The areas considered for

selection were those classified under Urban Redevelopment Authority Singapore postal districts 16, 20, 22 and 25, as these districts were near to the various study sites being set up across the country. Under this sampling frame, 15,000 household addresses were randomly

U

selected and notified of the survey by post. Subsequently, house visits were conducted to

N

enumerate all household members who met the following inclusion criteria: i) must be a

A

Singaporean or Permanent Resident (PR); ii) aged 18 to 79 years old; iii) stayed at least 4 days each week in the household and would be staying for the next 3 months or longer; iv)

M

not pregnant, having a severe mental illness, bedridden or wheelchair bound. Amongst those households that were successfully enumerated, one member per household was selected,

ED

using the Kish grid, to participate in the study. Assuming a LTBI prevalence of 4.3%, an estimated sample size of 1580 was planned in order to achieve a 95% level of confidence

PT

with a precision of 1%. Study procedures

CC E

An interviewer-administered electronic questionnaire was used to elicit information on the demographic, socio-economic, lifestyle practices relating to disease risk factors, medical history including past diagnosis of TB and possible exposure to TB. All study team members

A

were briefed extensively on the study methodology and underwent rigorous training to ensure consistency of and compliance to study procedures. Sample processing and laboratory analysis Blood samples collected for the QFT-GIT assay were sent to the Department of Laboratory Medicine at Tan Tock Seng Hospital and the National University Hospital Referral

4

Laboratory on the same day of collection and processed, interpreted and reported as per manufacturer’s instructions. Samples with indeterminate results were not repeated. Data management and analysis Responses recorded with the electronic questionnaire were automatically checked for missing values, data type errors and range sensibility. For quality control, 20% of the surveys were randomly selected and verified against audio-recording of the interviews. Data anomalies

IP T

were clarified through direct verification with the respondents whenever necessary. Laboratory test results were provided in electronic format by the laboratory and merged with

SC R

the rest of the electronic data, which was then subjected to a series of range, logic and

consistency checks. Asymptomatic participants with a positive IGRA result were considered as having LTBI.

U

The study sample was weighted to adjust for unequal probability of selection and differential response levels and to account for under-represented groups in the population. For the

N

household enumeration exercise of SH2, sample weights (WEE) comprise weights that were

A

computed based on the dwelling type and region of dwelling. For the study fieldwork of SH2,

M

sample weights (WSF) comprise weights that were computed based on age, gender and ethnicity. For the LTBI study, sample weights (W IGRA) for IGRA non-response were

ED

computed based on age, educational status and average monthly household income. Poststratification weights (WPS), to account for under-represented groups in the population, were computed based on ethnicity, age and gender, with reference to Singapore resident population

PT

as of mid-2014. The overall sample weights were the product of WEE, WSF, WIGRA and WPS. These weights were then applied to the sample to produce weighted estimates of LTBI

CC E

prevalence, which were also compared against the incidence rates of active TB cases notified for various age, gender and ethnic strata in 2014 [16]. To identify independent risk factors, adjusted hazard ratios were obtained using the multivariable modified Breslow-Cox Proportional Hazard Model. Because latent TB is a relatively common condition, odds ratios

A

are not a good approximation of prevalence rate ratios. We therefore modified Breslow–Cox Proportional Hazards regression using the method proposed by Lee & Chia, where the adjusted hazard ratios obtained are equivalent to prevalent rate ratios [17].

We also

performed further stratified analyses on the prevalence of LTBI, history of past diagnoses and exposure to a household member who had TB, by the participant’s country of birth. The country of birth was determined by a participant’s previous or current citizenship for 5

Singapore citizens and permanent residents respectively. Singaporean citizens who did not previously hold citizenship in other countries were considered as being of Singapore-born. Permanent residents and citizens who had previous citizenship elsewhere were considered to be foreign-born.

Ethical considerations

IP T

The study protocol and procedures were approved by the National University of Singapore

Institutional Review Board (NUS IRB reference no.:13-512), while written informed consent was obtained from each participant, as well as the parent or legal guardian for participants

SC R

aged 18 to 20 years.

U

RESULTS Characteristics of study population

N

Amongst the 2686 Singapore residents who participated in the SH2 study, 1690 (63%) of

A

them provided blood for IGRA (Table 1). A comparison between IGRA respondents and non-

M

respondents showed significant differences by age, educational status and monthly household income. After applying the SH2 weights to the IGRA respondents, about 20% each were in the first four age bands up to 59 years of age, 51.0% were female, and 76.0% were Chinese;

ED

these age, gender and ethnic distributions were fairly close to those of the Singapore resident population in 2014. However, the census-based estimates of the Singapore Resident

PT

Population had a larger proportion with primary or lesser education than our weighted distribution (30.4% versus 11.5% respectively), and a smaller proportion living in public

CC E

housing (82.6% versus 91.1% respectively). Factors associated with having a positive IGRA result. After excluding 8 indeterminate IGRA results, results from 1682 participants were left for

A

analysis. Of these, 213 (12.7%) had a positive IGRA result. Table 2 shows that the proportion positive by IGRA increased from 2.4% in those aged 18 to 29 years to 23.2% in those aged 70 to 79 years (p<0.001), and 10.6% of females were positive compared with 15.3% of males (p=0.004). Malays were least likely to have a positive IGRA result (7.5%), followed by Chinese (11.9%), Indians (18.6%) and those of other ethnic groups (20.6%, p=0.001). Singapore-born respondents were less likely to be positive than those are foreign-born

6

(10.7% versus 17.2%, p<0.001), and there were also significant differences (at p<0.001) by marital and educational status. In particular, those with the least education were more likely to be positive than those with post-secondary education (22.7% versus 10.3%). Those with self-reported past history of a TB diagnosis were significantly and much more likely to have positive IGRA results (70.4%, p<0.001). Those reporting a past exposure to a household member with TB were also significantly more likely to be positive than those who did not (21.0% versus 12.2%, p=0.021). Ex-smokers and current smokers were more likely to have a

IP T

positive result, as were those who consumed alcohol more than once a week and those who

had been diagnosed with diabetes. However, on multivariable regression, only some of these

SC R

factors remained positively and significantly associated with a positive IGRA at p<0.05,

namely older age, Indian ethnicity, being foreign-born, having primary or lesser education, self-reported past history of TB diagnosis, and consuming alcohol more than once a week. Those living in individual properties with privately owned land were less likely to be positive

N

U

by IGRA (adjusted Hazard Ratio 0.1, 95% Confidence Intervals (CI): 0.0 – 0.5, p=0.008).

A

Variation by country of birth

M

Amongst those foreign-born, the proportion positive by IGRA varied greatly by country of birth (Table 3). The highest prevalence of 30.8% was observed in those from India, followed by participants from Southeast Asian countries (27.0%) and those from China (17.1%), with

ED

the prevalence amongst Singapore-born (10.7%) being significantly lower (p<0.001 versus India and also other Southeast Asian countries, and p=0.043 for China). Participants from

PT

Malaysia had a prevalence that was slightly but not significantly lower (7.3%, p=0.140) and the small number from other countries had a prevalence that was slightly but not significantly

CC E

higher (11.5%, p=0.885).

Self-reported past history of a TB diagnosis was uncommon in those from Singapore (only 1.4%), and lower than those from other South East Asian countries (4.1%, p=0.100) and India

A

(4.7%, p=0.026). History of exposure to TB in a household member was reported by 5.5% of Singapore-born participants, which was not significantly different from that reported by foreign-born participants. In all, while 5.9% of respondents reported having either a past TB diagnosis or an exposure to TB, only 33 of these 100 respondents had a positive IGRA result. Moreover, these 33 individuals with past TB or exposure history represented only 15.5% of all the 213 respondents with a positive IGRA result. 7

Since China, India and most of the other South East Asian countries are classified by WHO as high burden countries [1] and contribute substantially to the population of naturalized citizens and permanent residents in Singapore, we grouped respondents from these countries together when performing additional age stratified analyses. As expected, there was an increasing trend in the proportion positive by IGRA with age, both amongst Singapore-born participants and those born in key high burden countries (Figure 1). Amongst those

IP T

Singapore-born, the prevalence increased from 2.0% (95% CI: 0.5 – 5.1) in those aged 18 to

29 years to 23.3% (95% CI: 13.4 – 36.0) in those aged 70 to 79 years. For those from key

SC R

high burden countries, the prevalence in the youngest age group was similar to those

Singapore-born, at 3.3% (95% CI: 0.1 – 17.2). However, there was a marked difference amongst respondents between the ages 30 to 49 years. For those from key high burden countries, LTBI prevalence was 20.8% (95% CI: 14.1 – 29.0) in those aged 30 to 39 years,

U

and 30.1% (95% CI: 21.0 – 40.5) in those aged 40 to 49 years; this was 3.2 and 4.0 times the

N

respective prevalence in Singapore-born participants (6.6%, 95% CI: 3.2 – 11.8; and 7.5%,

M

A

95% CI: 4.5 – 11.6).

Comparison of LTBI prevalence and incidence rates for active TB Lastly, we used our sampling weights to estimate the prevalence in various sociodemographic

ED

strata for which TB notification data are available (Figure 2). The overall weighted LTBI prevalence was 10.7% (95% CI: 9.0 – 12.4) when including all respondents but 9.2% (95%

PT

CI: 7.2 – 11.1) when restricted to Singapore-born respondents. Amongst all respondents, the weighted prevalence in females (8.8%, 95% CI: 6.7 – 11.0) was lower than in males (12.7%,

CC E

95% CI: 13.8 – 19.1). The same trend was also observed when only Singapore-born respondents were included in the analysis (6.4%, 95% CI: 4.0 – 8.8 versus 11.8%, 95% CI: 8.7 – 14.9). In contrast to results in Table 2 and the high weighted prevalence of LTBI in Indians amongst all respondents in Figure 2A (17.2%, 95% CI: 11.7 – 22.7), Singapore-born

A

Indians had an LTBI prevalence (7.0%, 95% CI: 2.3 – 11.7) similar to Malays (6.6%, 95% CI: 2.3 – 10.9) and lower than in Chinese (9.9%, 95% CI: 7.5 – 12.3%). Notably, although Malays had the lowest prevalence of LTBI, they had the highest incidence rates of active TB (about 1.6 times that for Chinese).

8

Estimates of LTBI prevalence were markedly higher when including all respondents versus only Singapore-born aged 30 to 39 (9.8% versus 6.3%) and those aged 40 to 49 years (11.5% versus 7.1%). In those aged 70 to 79, LTBI prevalence was higher when including only the Singapore-born (29.4%, 95% CI: 14.1 – 44.7 versus 26.0%, 95% CI: 12.5 – 39.4 when including all respondents). Both LTBI prevalence and age-stratified incidence rates of active

IP T

TB increased with age, with the two showing reasonable correlation across the age groups.

DISCUSSION

SC R

Our cross-sectional population survey found an overall IGRA (QFT-GIT) positivity rate of 12.7% in Singapore residents from which a weighted LTBI prevalence of 10.7% was derived.

There was a wide variation in IGRA positivity according to the residents’ country of birth,

U

reflecting the TB rates of these countries. This ranged from 7.3% and 10.7% in those born in Malaysia and Singapore respectively to 27% and 30% among residents born in South East

N

Asian countries and India respectively. Not unexpectedly, weighted LTBI prevalence

A

increased with age, with a prevalence of 29.4% among Singapore-born aged 70 to 79 years.

M

Higher LTBI prevalence was also not unexpectedly associated with lower educational and socio-economic status and alcohol use. The strength of this study was the large population

ED

surveyed with accredited laboratory testing. The interpretation of our findings should take into account the QFT GIT sensitivity of 83% in our local population using culture positive PTB as a surrogate [18] and its possible waning in persons who have been treated [19] or

PT

after years of infection [20]. The other study limitations were the low response rate to blood taking for the IGRA (only 63% of the study participants provided blood for QFT-GIT) and

CC E

the significant differences by age, educational and socioeconomic status between the IGRA respondents and non-respondents. We used sampling and post-stratification weights to account for differential sampling rates in major population sub-groups, and differential nonresponse rates. This standard approach to handling survey data should reduce bias and

A

improve the stability of estimates. However, selection bias may not be adequately removed even after weighting, especially if the variables available for weighting insufficiently account for the probability of non-response. Further, the derivation of weights in complicated designs such as in this study may not sufficiently correct for population representativeness. Interpretations of our findings need to take into account these limitations. Finally, although the household addresses were selected only from districts near the various study sites, we 9

made sure that the locations of the study sites were extensive and covered the west, north, north-east and south-east central regions of Singapore.

It was surprising that the Malays who have the highest TB incidence rate among the three main ethnic groups in Singapore, had the lowest LTBI prevalence among these groups. A study found TB patients of Malay ethnicity in Singapore to have significantly more infectious

IP T

(ie. sputum smear positive and cavitary) disease, and to be clustered at the same residential

address, implying increased transmission within their community [21]. Malay TB patients were also more likely to be diabetic. A possible reason for the unexpectedly low LTBI

SC R

prevalence in this group may be the lack of representativeness of the sample population.

Another possibility may be the difference in the IGRA performance between ethnic groups. A study comparing the T-SPOT.TB and QFT GIT in patients with culture positive PTB at the

U

Singapore TB Control Unit found IGRAs to be less sensitive in Malays and Indians,

N

compared to the Chinese population [18]. We may also hypothesize that this ethnic group may be genetically predisposed to a higher risk of progression to active disease once latently

M

A

infected. This would be worthy of investigation in future studies.

ED

To our knowledge, reports of population-wide surveys utilizing IGRAs have thus far been few. The United States (US) National Health and Nutrition Examination Survey (NHANES) 2011-2012 utilized TST and IGRA (QFT-GIT) to estimate the prevalence of TB infection

PT

[22]. Using the TST, they found no difference in the estimated prevalence of TST positivity rate from 1999-2000 (4.3%) to 2011-2012 (4.7%) despite a continued decline of TB disease

CC E

in the US over the same period. The IGRA positive rate in 2011-2012 was 5.0% and double TST and IGRA positivity was 2.1%. The point estimate of IGRA positivity prevalence in foreign-born persons was lower than that for the TST (15.9% versus 20.5%), possibly due to false TST positivity as a result of cross-reactivity with BCG. A population-based, multicentre

A

study in rural China which used both TST and QFT-GIT also showed higher TST positivity rates (15-42%) compared to QFT (13- 20%), suggesting that the prevalence of LTBI in China may be overestimated by the TST [23]. In contrast, a population-based survey in Saudi Arabia (where BCG is routinely given at birth) showed similar LTBI prevalence using TST (9.3%) and QFT-GIT (9.1%); with an annual risk of tuberculosis infection (ARTI) of 0.36% using TST and 0.35% using QFT-GIT [24]. A study utilizing TST and QFT-GIT to assess 10

ARTI among school children in Spain found a much lower ARTI calculated using the QFT (0.12%) than that using the TST (0.60%) [25].

Information on the background prevalence of LTBI in Singapore residents gleaned from this study provides evidence-based guidance to the STEP which has, since 1998, performed contact investigation to identify recently infected contacts who are at high risk for

IP T

progression to active disease and are candidates for preventive therapy [26]. Contact screening is carried out according to concentric circles of exposure to the infectious index case starting with the innermost circle, and the decision to expand screening based on the

SC R

attack rate of each circle [27]. Generally, no further expansion of screening would be required when the attack rate approaches the background LTBI rate of the community. Knowledge of the age-stratified LTBI prevalence in the community is thus useful to guide contact

N

U

investigations, particularly in schools and nursing homes.

A

Since 2005, there has been a marked increase in the number of foreign born residents and

M

non-residents from high TB incidence countries in Singapore, with a concomitant increase in the number of TB cases among this population such that foreign-born TB cases constitute

ED

more than 40% of the total TB burden in Singapore [11]. The high LTBI prevalence among residents from surrounding South East Asian countries and India gives a forecast of the future TB situation in Singapore. Some may argue for the screening of immigrants from high TB

PT

burden countries for LTBI treatment [8]. While this may have its merits in low TB incidence countries, the feasibility and cost-effectiveness of this intervention in Singapore should be

CC E

more carefully explored. Until less costly screening tools with higher positive predictive values than the current IGRAs and shorter preventive therapy regimens with negligible adverse effects become available, the country’s resources may be better utilized for higher priority TB control activities, namely early detection and treatment of active TB cases,

A

infection control and contact screening.

CONCLUSIONS This first LTBI prevalence study in Singapore sheds light on the reservoir of TB infection and the potential size of the problem of TB in the country. It provides useful information to 11

the national TB programme in its implementation of contact screening. Sequential surveys to determine the trend of prevalence of TB infection in the community should be performed to aid our on-going battle against this ancient disease of man.

FUNDING

IP T

Study was funded by the Ministry of Health, Singapore, under the CD-PHRG (Grant number: MOHCS14MAR001). The funder had no role in study design, data collection and analysis or preparation of the manuscript. The corresponding author has full access to all the data in the

SC R

study and final responsibility for the decision to submit for publication.

U

AUTHORS’ CONTRIBUTIONS

N

WYL, TB, MC, YTW and CC were integral to the study conception and development of the methodology. WYL, YTW, CC and LT were involved in the questionnaire design and

A

acquisition of the data. PY, KT, WYL, MC and CC participated in the analysis and

ED

revisions to the manuscript.

M

interpretation of the data. PY, KT, TB, MC, YTW and CC drafted and provided critical

PT

CONFLICT OF INTEREST

CC E

All authors declare no competing interests.

ACKNOWLEDGEMENTS We would like to thank the Singaporean residents who had taken time to participate in this

A

study.

12

REFERENCES

A

CC E

PT

ED

M

A

N

U

SC R

IP T

[1] World Health Organization, Global Tuberculosis report 2017, World Health Organization, Geneva, Switzerland, 2017. http://www.who.int/tb/publications/global_report/en/. [2] K. Dheda, T. Gumbo, N.R. Gandhi, M. Murray, G. Theron, Z. Udwadia, G.B. Migliori, R. Warren, Global control of tuberculosis: from extensively drug-resistant to untreatable tuberculosis, Lancet Respir Med. 2 (2014) 321–338. doi:10.1016/S2213-2600(14)700311. [3] World Health Organization, Guidelines on the management of latent tuberculosis infection, World Health Organization, Geneva, Switzerland, 2015. http://www.who.int/tb/publications/ltbi_document_page/en/ (accessed May 3, 2018). [4] US Centers for Disease Control and Prevention (CDC), The difference between latent TB infection and TB disease, (2014). https://www.cdc.gov/tb/publications/factsheets/general/ltbiandactivetb.htm (accessed May 3, 2018). [5] J.-W. Ai, Q.-L. Ruan, Q.-H. Liu, W.-H. Zhang, Updates on the risk factors for latent tuberculosis reactivation and their managements, Emerg Microbes Infect. 5 (2016) e10. doi:10.1038/emi.2016.10. [6] R.M.G.J. Houben, P.J. Dodd, The Global Burden of Latent Tuberculosis Infection: A Reestimation Using Mathematical Modelling, PLoS Med. 13 (2016) e1002152. doi:10.1371/journal.pmed.1002152. [7] H. Alsdurf, P.C. Hill, A. Matteelli, H. Getahun, D. Menzies, The cascade of care in diagnosis and treatment of latent tuberculosis infection: a systematic review and metaanalysis, Lancet Infect Dis. 16 (2016) 1269–1278. doi:10.1016/S1473-3099(16)30216-X. [8] H. Getahun, A. Matteelli, I. Abubakar, M.A. Aziz, A. Baddeley, D. Barreira, S. Den Boon, S.M. Borroto Gutierrez, J. Bruchfeld, E. Burhan, S. Cavalcante, R. Cedillos, R. Chaisson, C.B.-E. Chee, L. Chesire, E. Corbett, M. Dara, J. Denholm, G. de Vries, D. Falzon, N. Ford, M. Gale-Rowe, C. Gilpin, E. Girardi, U.-Y. Go, D. Govindasamy, A. D Grant, M. Grzemska, R. Harris, C.R. Horsburgh, A. Ismayilov, E. Jaramillo, S. Kik, K. Kranzer, C. Lienhardt, P. LoBue, K. Lönnroth, G. Marks, D. Menzies, G.B. Migliori, D. Mosca, Y.D. Mukadi, A. Mwinga, L. Nelson, N. Nishikiori, A. Oordt-Speets, M.X. Rangaka, A. Reis, L. Rotz, A. Sandgren, M. Sañé Schepisi, H.J. Schünemann, S.K. Sharma, G. Sotgiu, H.R. Stagg, T.R. Sterling, T. Tayeb, M. Uplekar, M.J. van der Werf, W. Vandevelde, F. van Kessel, A. van’t Hoog, J.K. Varma, N. Vezhnina, C. Voniatis, M. Vonk Noordegraaf-Schouten, D. Weil, K. Weyer, R.J. Wilkinson, T. Yoshiyama, J.P. Zellweger, M. Raviglione, Management of latent Mycobacterium tuberculosis infection: WHO guidelines for low tuberculosis burden countries, Eur. Respir. J. 46 (2015) 1563– 1576. doi:10.1183/13993003.01245-2015. [9] Singapore Department of Statistics, Latest data - Population & land area (mid-year estimates), (2016). https://www.singstat.gov.sg/statistics/latest-data#16 (accessed May 3, 2018). [10] C.B.-E. Chee, Y.T. Wang, TB control in Singapore: where do we go from here?, Singapore Med J. 53 (2012) 236–238. [11] Ministry of Health Singapore, Communicable diseases surveillance in Singapore 2017, 2017. https://www.moh.gov.sg/content/moh_web/home/Publications/Reports/2017/communica ble-diseases-surveillance-in-singapore-2016.html.

13

A

CC E

PT

ED

M

A

N

U

SC R

IP T

[12] W. Wah, S. Das, A. Earnest, L.K.Y. Lim, C.B.E. Chee, A.R. Cook, Y.T. Wang, K.M.K. Win, M.E.H. Ong, L.Y. Hsu, Time series analysis of demographic and temporal trends of tuberculosis in Singapore, BMC Public Health. 14 (2014) 1121. doi:10.1186/1471-2458-14-1121. [13] L.C. Kahwati, C. Feltner, M. Halpern, C.L. Woodell, E. Boland, H.R. Amick, R.P. Weber, D.E. Jonas, Primary Care Screening and Treatment for Latent Tuberculosis Infection in Adults: Evidence Report and Systematic Review for the US Preventive Services Task Force, JAMA. 316 (2016) 970–983. doi:10.1001/jama.2016.10357. [14] C.B.E. Chee, L.K.Y. Lim, T.M. Barkham, D.R. Koh, S.O. Lam, L. Shen, Y.T. Wang, Use of a T cell interferon-gamma release assay to evaluate tuberculosis risk in newly qualified physicians in Singapore healthcare institutions, Infect Control Hosp Epidemiol. 30 (2009) 870–875. doi:10.1086/599284. [15] R. Menzies, B. Vissandjee, Effect of bacille Calmette-Guérin vaccination on tuberculin reactivity, Am. Rev. Respir. Dis. 145 (1992) 621–625. doi:10.1164/ajrccm/145.3.621. [16] Ministry of Health Singapore, Communicable diseases surveillance in Singapore 2014, 2015. https://www.moh.gov.sg/content/moh_web/home/Publications/Reports/2015.html. [17] J. Lee, K.S. Chia, Estimation of prevalence rate ratios for cross sectional data: an example in occupational epidemiology, Br J Ind Med. 50 (1993) 861–862. [18] C.B.E. Chee, S.H. Gan, K.W. Khinmar, T.M. Barkham, C.K. Koh, S. Liang, Y.T. Wang, Comparison of sensitivities of two commercial gamma interferon release assays for pulmonary tuberculosis, J. Clin. Microbiol. 46 (2008) 1935–1940. doi:10.1128/JCM.02403-07. [19] C.B.E. Chee, K.W. KhinMar, S.H. Gan, T.M. Barkham, C.K. Koh, L. Shen, Y.T. Wang, Tuberculosis treatment effect on T-cell interferon-gamma responses to Mycobacterium tuberculosis-specific antigens, Eur. Respir. J. 36 (2010) 355–361. doi:10.1183/09031936.00151309. [20] T. Mori, N. Harada, K. Higuchi, Y. Sekiya, K. Uchimura, T. Shimao, Waning of the specific interferon-gamma response after years of tuberculosis infection, Int. J. Tuberc. Lung Dis. 11 (2007) 1021–1025. [21] L.K.-Y. Lim, D.A. Enarson, A.J. Reid, S. Satyanarayana, J. Cutter, K.M. Kyi Win, C.B.-E. Chee, Y.T. Wang, Notified tuberculosis among Singapore residents by ethnicity, 2002-2011, Public Health Action. 3 (2013) 311–316. doi:10.5588/pha.13.0055. [22] R. Miramontes, A.N. Hill, R.S. Yelk Woodruff, L.A. Lambert, T.R. Navin, K.G. Castro, P.A. LoBue, Tuberculosis Infection in the United States: Prevalence Estimates from the National Health and Nutrition Examination Survey, 2011-2012, PLoS ONE. 10 (2015) e0140881. doi:10.1371/journal.pone.0140881. [23] L. Gao, W. Lu, L. Bai, X. Wang, J. Xu, A. Catanzaro, V. Cárdenas, X. Li, Y. Yang, J. Du, H. Sui, Y. Xia, M. Li, B. Feng, Z. Li, H. Xin, R. Zhao, J. Liu, S. Pan, F. Shen, J. He, S. Yang, H. Si, Y. Wang, Z. Xu, Y. Tan, T. Chen, W. Xu, H. Peng, Z. Wang, T. Zhu, F. Zhou, H. Liu, Y. Zhao, S. Cheng, Q. Jin, LATENTTB-NSTM study team, Latent tuberculosis infection in rural China: baseline results of a population-based, multicentre, prospective cohort study, Lancet Infect Dis. 15 (2015) 310–319. doi:10.1016/S14733099(14)71085-0. [24] H.H. Balkhy, K. El Beltagy, A. El-Saed, B. Aljasir, A. Althaqafi, A.F. Alothman, M. Alshalaan, H. Al-Jahdali, Prevalence of Latent Mycobacterium Tuberculosis Infection (LTBI) in Saudi Arabia; Population based survey, Int. J. Infect. Dis. 60 (2017) 11–16. doi:10.1016/j.ijid.2017.03.024.

14

A

CC E

PT

ED

M

A

N

U

SC R

IP T

[25] N. Díez, E. Giner, I. Latorre, A. Lacoma, F.-J. Roig, I. Mialdea, J. Díaz, M. SerraVidal, A. Escribano, J. Domínguez, Use of interferon-gamma release assays to calculate the annual risk of tuberculosis infection, Pediatr. Infect. Dis. J. 34 (2015) 219–221. doi:10.1097/INF.0000000000000514. [26] C.B.E. Chee, M.D. Teleman, I.C. Boudville, S.E. Do, Y.T. Wang, Treatment of latent TB infection for close contacts as a complementary TB control strategy in Singapore, Int. J. Tuberc. Lung Dis. 8 (2004) 226–231. [27] J. Veen, Microepidemics of tuberculosis: the stone-in-the-pond principle, Tuber. Lung Dis. 73 (1992) 73–76. doi:10.1016/0962-8479(92)90058-R.

15

FIGURES Figure 1 Comparison of LTBI prevalence by age groups in participants born in “Singapore” (green diamonds) against those from “Key high burden countries” (orange triangles) and “Other countries” (blue circles). Key high burden countries include China, India and other South East Asian countries besides Malaysia (which is included under Other countries).

80% 70%

Proportion with positive IGRA by country of origin Singapore Key high burden countries Other countries

60% 50%

U

40% 30%

N

20% 10%

A

% with positive IGRA result

90%

SC R

100%

IP T

(Note: Error bars denote 95% confidence intervals.)

0%

40-49 50-59 Age group, yrs

M

30-39

60-69

70-79

A

CC E

PT

ED

18-29

16

Figure 2 Weighted LTBI in comparison to notified cases of active TB per 100,000 population (blue bars) in citizens and Singapore Permanent Residents by (A) gender and ethnicity and (B) age group. Orange circles and green diamonds give the prevalence estimated with the sample weights for all respondents and just the Singapore-born respectively.

A

CC E

PT

ED

M

A

N

U

SC R

IP T

(Note: Error bars denote 95% confidence intervals. The group aged <30 years for LTBI prevalence includes participants aged 18 – 29 years while the TB notification data was based on those aged 20 – 29 years.)

17

TABLES Table 1 Comparison of the socio-demographic and socio-economic profiles between the IGRA respondents and non-respondents, and between the weighted sample of IGRA respondents and the Singapore Resident Population of 2014 (N=2,686)

Characteristics

IGRA Respondents, N=1690

pvalue†

Singapore IGRA Respon- Resident dents, weighted Popula% (95% CI) tion 2014, %

IP T

No. (%) for IGRA Non-respondents, N=996

A

CC E

PT

ED

M

A

N

U

SC R

Age group, yrs 18–29 187 (18.8) 245 (14.5) 20.6 (18, 23.3) 24.5 30–39 211 (21.2) 329 (19.5) 19.7 (17.5, 22) 18.7 40–49 184 (18.5) 422 (25.0) 20.7 (18.6, 22.9) 19.7 <0.001 50–59 161 (16.2) 355 (21.0) 20.1 (17.8, 22.4) 19.0 60–69 165 (16.6) 268 (15.9) 12.9 (10.9, 15) 12.4 70– 79 88 (8.8) 71 (4.2) 5.9 (4.2, 7.5) 5.8 Gender Female 539 (54.1) 941 (55.7) 51.0 (48.0, 53.9) 50.9 0.431 Male 457 (45.9) 749 (44.3) 49.0 (46.1, 52.0) 49.1 Ethnicity Chinese 644 (64.7) 1146 (67.8) 76.0 (73.7, 78.3) 75.3 Malay 164 (16.5) 226 (13.4) 12.4 (10.6, 14.2) 12.8 0.152 Indian 154 (15.5) 255 (15.1) 8.8 (7.5, 10.1) 8.8 Others 34 (3.4) 63 (3.7) 2.9 (2.0, 3.7) 3.1 Country of birth Singapore 726 (72.9) 1171 (69.3) 72.3 (69.7, 74.8) Not 0.048 Non-Singapore 270 (27.1) 519 (30.7) 27.7 (25.2, 30.3) available Educational Status Post-secondary 456 (45.8) 874 (51.7) 58.7 (55.8, 61.5) 51.2 Secondary 339 (34.0) 571 (33.8) <0.001 29.8 (27.2, 32.4) 18.4 Primary or less 201 (20.2) 245 (14.5) 11.5 (9.8, 13.3) 30.4 Dwelling Type HDB Flats (Public housing) 955 (95.9) 1629 (96.4) 91.1 (88.6, 93.6) 82.6 0.796 Condominium 19 (1.9) 29 (1.7) 5.8 (3.6, 8.0) 10.8 Landed Property 22 (2.2) 32 (1.9) 3.1 (1.7, 4.5) 6.6 Monthly Household Income (SGD)‡ <$2000 265 (30.1) 370 (23.8) 17.8 (15.6, 19.9) $2000 –$3999 231 (26.3) 457 (29.4) 28.8 (26, 31.5) Not $4000–$5999 157 (17.8) 324 (20.8) 0.010 22.5 (19.9, 25.1) available $6000–$9999 145 (16.5) 263 (16.9) 19.8 (17.3, 22.3) ≥$10000 82 (9.3) 141 (9.1) 11.1 (8.9, 13.4) †Chi-squared test comparing distribution of IGRA respondents and non-respondents ‡Excludes 251 respondents who did not provide a valid response

18

No. (%)positi ve for pLTBI value†

Multivariable analysis

246 327 422 349 265 73

6 (2.4) 38 (11.6) 51 (12.1) 57 (16.1) 45 (17.0) 16 (23.2)

Referent 3.4 (1.2, 10.0) 3.5 (1.2, 10.2) 4.3 (1.4, 13.1) 3.9 (1.2, 12.0) 7.2 (2.0, 25.5)

0.024 0.019 0.010 0.020 0.002

936 746

99 (10.6) 0.004 114 (15.3)

Referent 1.4 (1.0, 2.0)

0.049

1,140 226 253 63

136 (11.9) 0.001 17 (7.5) 47 (18.6) 13 (20.6)

Referent 0.7 (0.4, 1.3) 1.6 (1.1, 2.2) 1.2 (0.6, 2.3)

0.290 0.015 0.662

1,164 518

124 (10.7) <0.001 89 (17.2)

Referent 1.6 (1.1, 2.2)

0.007

385 1,104 130 63

25 (6.5) <0.001 153 (13.9) 23 (17.7) 12 (19.0)

Referent 0.9 (0.5, 1.6) 1.2 (0.6, 2.6) 0.8 (0.3, 1.9)

0.755 0.620 0.561

242 568 872

55 (22.7) 68 (12.0) 90 (10.3)

Referent 1.1 (0.7, 1.8) 2.1 (1.3, 3.6)

0.554 0.004

1,621 29 32

208 (12.8) 0.258 4 (13.8) 1 (3.1)

Referent 0.7 (0.2, 1.9) 0.1 (0.0, 0.5)

0.452 0.008

368 455 323 263 140 133

57 (15.5) 57 (12.5) 41 (12.7) 24 (9.1) 15 (10.7) 19 (14.3)

Referent 1.5 (0.9, 2.4) 1.3 (0.8, 2.2) 1.1 (0.6, 1.9) 1.3 (0.7, 2.7) 1.8 (1.0, 3.1)

0.098 0.259 0.852 0.422 0.053

1,655 27

194 (11.7) <0.001 19 (70.4)

Referent 5.8 (3.7, 9.0)

<0.001

1,601 81

196 (12.2) 0.021 17 (21.0)

Referent 1.1 (0.6, 2.1)

0.699

1,298

148 (11.4) 0.015

Referent

A

CC E

PT

IP T

Adj HR‡ (95% CI) p-value

N

U

SC R

<0.001

M

ED

Characteristics Age group, years 18 – 29 30 – 39 40 – 49 50 – 59 60 – 69 70 – 79 Gender Female Male Ethnicity Chinese Malay Indian Others Country of birth Singapore Non-Singapore Marital Status Never married / Not stated Currently married Separated/Divorced Widowed Educational Status Post-secondary Secondary Primary or less Dwelling Type HDB Flats (Public housing) Condominium Landed Property Monthly Household Income (SGD) < $2000 $2000 – $3999 $4000 – $5999 $6000 – $9999 ≥$10000 Not stated Past history of TB diagnosis No Yes Exposed to TB in household member No Yes Smoking Status Never a smoker

Total respondents

A

Table 2 No. of respondents with positive IGRA result and adjusted Hazard Ratios (adj HR) for a positive IGRA result by participant characteristics (N=1682)

<0.001

0.266

19

A

CC E

PT

ED

M

A

N

U

SC R

IP T

Ex-smoker 144 26 (18.1) 1.2 (0.7, 2.0) 0.533 Current occasional smoker 42 4 (9.5) 1.0 (0.4, 2.8) 0.990 Current daily smoker 198 35 (17.7) 1.3 (0.8, 2.0) 0.344 Alcohol Consumption None 855 111 (13.0) <0.001 Referent Less than once per month 487 51 (10.5) 1.0 (0.7, 1.5) 0.864 Several days per month 178 15 (8.4) 0.9 (0.5, 1.5) 0.647 More than once per week 162 36 (22.2) 1.8 (1.2, 2.8) 0.009 Diagnosed with Diabetes No 1,546 180 (11.6) <0.001 Referent Yes 136 33 (24.3) 1.4 (0.9, 2.2) 0.128 †Chi-squared test comparing distribution for those positive and those negative for LTBI ‡Adjusted Hazard Ratios obtained using multivariable modified Breslow-Cox Proportional Hazard Model with robust variance with sampling weights and robust variance, and including all the variables in the above table

20

Table 3 Past history of TB diagnosis, contact with TB and proportion with positive IGRA result by country of birth (N=1,682) Total respondents, N=16 82†

Country of birth

No. (%) of respondents with:

No. (%) of positive Past Exposed Either IGRA history to TB in history of with past of TB houseTB or TB or diaghold exposure exposure nosis member to TB history‽ 27 (1.6) 81 (4.8) 100 (5.9) 33 (15.5) 16 (1.4) 64 (5.5) 74 (6.4) 20 (16.1) 11 (2.1) 17 (3.3) 26 (5.0) 13 (14.6) 1 (0.5) 5 (2.4) 6 (2.9) 1 (6.7) 3 (4.1) 3 (4.1) 5 (6.8) 3 (15.0) 1 (1.0) 6 (5.7) 7 (6.7) 2 (11.1) 5 (4.7) 3 (2.8) 7 (6.5) 6 (18.2) 1 (3.8) 0 (0.0) 1 (3.8) 1 (33.3)

LTBI

A

CC E

PT

ED

M

A

N

U

SC R

IP T

All respondents 1682 213 (12.7) Singapore 1164 124 (10.7) Non-Singapore 518 89 (17.2) Malaysia 206 15 (7.3) Other South East Asia‡ 74 20 (27.0) China 105 18 (17.1) India 107 33 (30.8) Other countries 26 3 (11.5) Key high burden 286 71 (24.8) 9 (3.1) 12 (4.2) 19 (6.6) 11 (15.5) countries₰ †Excludes 8 participants with indeterminate IGRA result ‽Denominator for the percentage here is the number with positive IGRA result ‡Includes Indonesia, Thailand, Philippines, Myanmar, Vietnam, Cambodia, Laos and Brunei ₰Other South East Asia, China or India

21