Cheng et al.

Infectious Diseases of Poverty (2020) 9:13

https://doi.org/10.1186/s40249-019-0614-9

RESEARCH ARTICLE Open Access

Incidence and risk factors of tuberculosis

among the elderly population in China: a
prospective cohort study
Jun Cheng1†, Yan-Ni Sun2†, Can-You Zhang1†, Yan-Ling Yu3†, Li-Hong Tang4†, Hong Peng5†, Ying Peng6†,
Yu-Xia Yao7†, Shuang-Yi Hou8†, Jian-Wei Li9†, Jin-Ming Zhao10†, Lan Xia11†, Lin Xu12†, Yin-Yin Xia1, Fei Zhao1,
Li-Xia Wang1* and Hui Zhang1*

Abstract
Background: China is facing challenges of the shifting presentation of tuberculosis (TB) from younger to elderly
due to an ageing population, longer life expectancy and reactivation disease. However, the burden of elderly TB
and influence factors are not yet clear. To fill the gap, we generated a cohort study to measure the magnitude of
TB incidence and associated factors among the elderly population aged 65 years and above in China.
Methods: In this cohort established in 2013 through a prevalence survey conducted in selected sites, a total of 34
076 elderlies without TB were enrolled into two-year follow-up. We used both active and passive case findings to
find out all TB patients among them. The person-year (PY) incidence rates for both bacteriologically positive TB and
active TB were calculated. Cox proportional regression model was performed to test effect of risk factors, and the
population attributable fraction (PAF) of each risk factor contributing to incident TB among elderlies was calculated.
Results: Over the two-year follow-up period, a total of 215 incident active TB were identified, 62 of which were
bacteriologically positive. The incidence rates for active TB and bacteriologically positive TB were 481.8 per 100 000
PY (95% CI: 417.4–546.2 per 100 000 PY) and 138.9 per 100 000 PY (95% CI: 104.4–173.5 per 100 000 PY), respectively.
Incident cases detected by active case finding were significantly higher (P < 0.001). Male, non-Han nationality,
previously treated TB, ex/current smoker and body mass index (BMI) < 18.5 presented as independent predictors for
developing TB disease. For developing bacteriologically positive TB, the biggest contribution was from self-reported
ex or current smoker (18.06%). And, for developing active TB, the biggest contribution was from non-Han
nationality (35.40%), followed by male (26.80%) and age at 75 years and above (10.85%).
Conclusions: Ageing population in China had a high TB incidence rate and risk to develop TB disease, implying
that National TB Program (NTP) needs to prioritize for elderly. Active case finding should be applied capture more
active TB cases among this particular population, especially for male, non-Han nationality, and those with identified
risk factors.
Keywords: Tuberculosis, Elderly tuberculosis, Active case finding, Incidence, Risk factor, Follow up, China

* Correspondence: wanglx@chinacdc.cn; zhanghui@chinacdc.cn

†
Jun Cheng, Yan-Ni Sun, Can-You Zhang are first authors and contributed
equally to this work. Yan-Ling Yu, Li-Hong Tang, Hong Peng, Ying Peng, Yu-
Xia Yao, Shuang-Yi Hou, Jian-Wei Li, Jin-Ming Zhao, Lan Xia, Lin Xu
Contributed equally to this work.
1
National Center for Tuberculosis Control and Prevention, Chinese Center for
Disease Control and Prevention, Beijing, China
Full list of author information is available at the end of the article

© The Author(s). 2020 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 2 of 13

Background in nine counties, by reviewing medical records at the par-

Increasing age raises the risk of Tuberculosis (TB). With ticipating health facilities and county-level social insurance
the increasing longevity and declining fertility rates, the system databases, reported that nearly 20% of TB patients
global pace of population ageing is getting faster in the were not reported to TBIMS, being higher percentage for
new century [1], and TB in the older adults is becoming children with TB, TB pleurisy, patients diagnosed in the
clinically important in many countries [2]. It is reported eastern and central regions and patients with a TB diagno-
by the World Health Organization (WHO) that TB notifi- sis recorded in either health facilities or social insurance
cation rates in older people (aged 65 years old and above) system [14, 15].
was higher than other 10-years interval groups in 2016 Reported TB incidence rate for each age group could be
[3], and many countries witness that the proportion of obtained from IDRS, and the reported TB incidence for
older TB patients remains high and has even increased in elderly was two to three times of that for younger adults,
some areas [4], with an increase disproportionately greater being similar difference on TB prevalence rates between
than the rise in the number of older persons. these two groups, obtained from the Fifth National Tuber-
China, a country with the second large TB burden in culosis Prevalence Survey [8]. However, this reported inci-
the world, is challenged by the shifting of TB from youn- dence underestimated the real status for elderly, because
ger to elderly due to the ageing population, longer life of both the missed cases by surveillance system itself and
expectancy and reactivation disease [5–7]. WHO esti- undiagnosed TB cases resulted from a high percent of
mated that the percentage of people aged 60 years or asymptomatic TB patients [16]. We established an elderly
over in China is expected to increase from 12.4% (168 incidence cohort without TB by conducting a prevalence
million people) in 2010 to 28.0% (402 million) in 2040 survey in selected sites in 2013, and utilized both active
[6]. In China, the fifth national TB prevalence survey and passive case findings to capture all incident TB pa-
conducted in 2010 showed the TB prevalence increased tients during two-year follow-up period, aiming to obtain
with age and peaked in the 75–79 age group, hitting the real TB incidence rate and identify risk factors for de-
866/100 000 [5, 8]. According to estimation by Huynh veloping TB among elderly.
et al. [9], China is unlikely to achieve the incidence tar-
get if the National Tuberculosis Program NTP failed to Methods
reduce TB incidence among elderlies. Significant impacts Cohort establishment from TB prevalence survey in 2013
of elderly TB control on TB programs had already been We conducted a prevalence survey in selected sites from
observed in Japan, and also in China [10]. If China aims July to September in 2013, to establish an elderly cohort
to achieve the milestone of End TB Strategy, the country without TB. The study sites selection and diagnostic algo-
has to strategically prioritize TB control among the el- rithm in that prevalence survey has been described fully
derlies. There are several approaches recommended to elsewhere [17]. In brief, the study adopted a three-stage
address elderly TB control in China and the choice cluster and random sampling for selecting study sites
should be based on TB burden among the elderlies [11]. (Fig. 1), and finally 10 townships and 17 communities lo-
TB epidemic in this high risk population should be cated in 10 counties covering 38 888 elderly were selected.
understood and is very helpful for policy making. For each elderly, door-to-door interview for data collec-
Although there are already TB surveillance systems in tion, physical examination for calculating body mass index
China, including Infectious Disease Reporting System (BMI), and chest X-ray (CXR) examination were provided
(IDRS) and TB Information Management System for them. Those elderlies who had suspected TB symptom
(TBIMS), the real TB incidence rate can not be obtained or abnormal chest radiography were required to provide
directly from them. IDRS, an extensive web-based and three sputum specimens (one “at spot”, one in early morn-
real-time surveillance system, enables all health facilities ing and one at night) for laboratory diagnosis. Finally,
across the country to report information on the 37 notifi- 4619 refused to participate this prevalence survey, leaving
able diseases (including TB) within 24 h [8]. And, the 34 269 elderlies finishing all required procedure, and 193
users of IDRS comprise all TB health facilities, including active TB cases were identified.
TB dispensaries and designated hospitals, at province, pre- The information obtained from door-to-door interview,
fecture and county level [12]. However, TBIMS do not including demographic, most of disease history, lifestyle
cover non-designated TB facilities, and all information and suspected TB symptoms, were totally self-reported by
covered by this system are about diagnosed TB patients participants, except for history of diabetes and treated TB
[13]. A special module in TBIMS acts as an interface be- previously. For diabetes information, we also checked their
tween TBIMS and IDRS, and information for one case can personal health document covered by the National Basic
be simultaneously updated in these two systems. However, Public Health Service Project and updated by year. Partici-
under-reporting of diagnosed TB has also existed in China pants self-reporting diabetes history or with a medical rec-
[14]. A recent retrospective inventory studies conducted ord for diabetes diagnosis were classified as known
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 3 of 13

Fig. 1 Sampling procedure of the tuberculosis prevalence survey in 2013

diabetes patients. For retreated TB patients, we checked survey), continuously lived in the village or community for
TBIMS to identify whether participants were diagnosed six-month and longer were eligible for our prospective
with TB and finished treatment or cured, and either self- study. Excluding 193 active TB identified in baseline
reporting or recorded as a treated TB patient previously prevalence survey, 34 076 elderlies without TB were en-
was classified as retreated TB cases. rolled into our incidence study.
 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
Sample size and study population Zα 2pð1−pÞ þ Z β p1 ð1−p1 Þ þ p2 ð1−p2 Þ
The sample size was calculated for a population cohort N¼
ðp1 −p2 Þ2
study. TB incidence rates for general population with TB
history and without TB history were 2706.3 per 100 000
person-year (PY) and 39.6 per 100 000 PY, respectively, Field procedure
obtained from another cohort study in China [18]. Assum- Our study population enrolled at baseline was set as a
ing TB incidence rate among elderly being two times of fixed cohort during this prospective study period, that is,
that for general population, we used 5412.6/100 000 only elderly enrolled into our incidence study at baseline
person-year and 79.2/100 000 person-year as the esti- will be followed up, and, those uncovered at baseline
mated TB incidence for elderly with TB history (P1) and and met inclusion criteria during the two-year follow-up
that for elderly without TB history (P2). The study was set period were excluded from our analysis.
at a power of 0.8 (one-side) and two-tailed significance of The informed consent was obtained from all partici-
0.05. Based on the formula, a total of 146 elderly with TB pants before the door-to-door interview at baseline sur-
history should be recruited. Assuming the percentage of vey. For elderlies with difficulties in communication, his/
persons with TB history in elderly was 20% higher than her family members were interviewed as the respondent.
that in general population (0.42% obtained from cohort During the two-year follow-up period, for those who
study in China [18]), the number of elderly needed to reg- were diagnosed as active TB cases, the information of
istered into cohort should be 28 765. Considering a 15% TB occurrence and date were immediately recorded, and
non-response rate and lost to follow up, we increased the further identified in TBIMS. And, every year for the
sample size to 33 080. In the 27 study sites, people who follow-up period, we conducted a door-to-door investi-
were born before September 31, 1948 (aged 65 years and gation to identify the follow-up status for each partici-
older at the time of implementing baseline prevalence pant, and promote them to participate required chest
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 4 of 13

radiography examination. Meanwhile, the information of Quality control

other outcomes, including death, transfer out or move, Strict quality control was performed across the whole study
and the date of outcome occurrence were asked and re- process. The questionnaire interview was carried out
corded accurately. among all eligible elderly face to face by trained staff at
From October 2013 to September 2015, we conducted county level. Questionnaires collected were checked by data
both active and passive case finding annually to detect and management specialist for completeness and logicality.
diagnose incident TB patients among this study cohort. After completing data entry, 2% of questionnaires were ran-
For active case finding, the same procedure and diagnostic domly selected for consistency check by the national expert
algorithm as baseline prevalence survey were used, that is, group. If discrepancies identified, information was revised
TB symptom screening and chest radiography examin- on the second day by asking the responsible interviewer. If
ation were provided for each sampled elderly, and sputum the interviewer could not work out the discrepancy, en-
smear microscopy and culture were performed for those quires were made to local staff to check with the respond-
with TB symptoms or abnormal chest radiography. For ent. All chest radiography films and sputum specimens
passive case finding, we asked detailed TB symptoms in- collected were transported to TB designated hospital at
formation among elderlies who actively visited township county level for diagnosis confirmation and strain identifi-
hospitals and village clinics during the study period. CXR cation. All chest radiography films for TB cases were
and collection of sputum specimen were administered to reviewed by a national expert group within two months
those with TB suspected symptoms. If there were any TB after yearly follow-up completion. The smear positive TB
cases reported during the study period, we collected their was assessed and classified strictly by trained laboratory
diagnosis and date. For those being diagnosed as TB cases technicians at county level according to the national stan-
by active case finding or passive case finding, treatment dardized methods and procedure. The quality of culture
was arranged by local hospitals. conducted in county laboratory was supervised and moni-
tored by prefectural or provincial TB laboratory. All the
sputum smear and culture results were checked by the staff
Study outcome and definition from the National Tuberculosis Reference Laboratory for
We measured the study outcomes as incident active TB, quality control. In addition to the rigorous internal quality
death, transfer and moving out from study sites, and refusal control for data collection, this study also had an independ-
to receive yearly chest radiography during the follow-up ent site auditing team from WESTAT (Rockville, MD,
period. We observed and recorded whichever came first as USA). The field inspectors from WESTAT went to study
an endpoint in the study according to the door-to-door sites regularly to check if the procedure performed followed
interview, passive case finding and yearly screening. the research legislation in terms of consent, progress, stor-
Active TB included bacteriologically positive TB and age and maintenance of records by working with the re-
clinical diagnosed TB. A bacteriologically positive TB pa- search team in the field to ensure study procedure being in
tient was defined as an individual with at least one sputum line with the study design and guarantee the data quality in
smear or culture positive. Smear positive was that the spu- terms of completion and authenticity.
tum specimen had at least one acid-fast bacillus identified
within 100 fields under microscopy. Culture positive was
that the specimen had at least one colony of Mycobacter- Data management and statistical analysis
ium tuberculosis complex being isolated by using Löwen- We developed a web-based dataset for data entry and
stein-Jensen medium [19]. The clinical diagnosed TB was management. All data collected from baseline and follow-
defined by the presence of three negative smears and CXR up surveys were double entered. Data clean and
abnormalities consistent with active pulmonary TB, and consistency check were performed by the national study
meeting at least one of the following criteria: clinical TB team. Statistical inference followed standard procedure for
symptoms, or strong positive purified protein derivative data analysis. Pearson’s χ2 test or Fisher’s exact test was
reaction, or TB lesions confirmed by histopathological used to test for differences in proportions. Two-sided P-
examination of extrapulmonary tissue, or excluding other values of < 0.05 were considered significant. The TB inci-
pulmonary disease by diagnostic treatment or follow-up dence rates for bacteriologically positive and active TB
observations [19]. The elderly who moved out and did not were calculated and expressed as cases per 100 000 PY.
come back during the study period was recorded as trans- The follow-up period, from the date of enrolment to the
ferred or moved out. Refusal was defined as an elderly occurrence of incident TB, death, transfer or moving out,
who refused to provide information, or to finish all re- completion of follow-up or the end of the study, was com-
quested TB check procedure. Moved or transferred out puted as the time passed. PY for each participant in this
and refusal were categorized as lost-to-follow-up in ana- cohort was calculated based on individual information,
lysis of this study. that is, the interval of finishing baseline survey date and
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 5 of 13

the exact date of outcome occurrence was calculated one Ethical Approval
by one and considered as his/her PY. The study protocol was approved by the Research Ethics
Univariate cox proportional regression model was per- Review Committee of Chinese Center for Disease
formed with time dependent covariate in relation to TB in- Control and Prevention (Approval number: 201322).
cidence using a forward inclusion approach. Demographic The signed informed consent was obtained from all par-
character, yearly household income per head, history of dis- ticipants before investigation.
eases, smoking and drinking status, and BMI at baseline
were regarded as influence factor and included into our
Results
analysis. We classified those with less than RMB 2300 in-
Elderlies followed up between 2013 and 2015
come per year as low-income group according to the na-
Among the 34 269 elderly people finishing baseline preva-
tional poor standard (2011), and for those with more
lence survey, 193 were detected as active TB cases, leaving
income, we divided them into two groups by cutoff point
34 076 elderlies in the two-year follow-up. At the end of
being RMB 10 000. According to disease history, partici-
follow-up period, 22 119 elderly people remained in the co-
pants was classified into yes or no or unknown groups. We
hort for data analysis (Including those transferred out or
categorized BMI as underweight (BMI < 18.5 kg/m2), nor-
moved, or refused follow-up check in the first follow-up
mal weight (≥ 18.5 kg/m2 and < 24.0 kg/m2), overweight
year, and finished the second year follow-up check) (Fig. 2).
and obese (≥ 24.0 kg/m2) [20]. Variables significantly corre-
lated in the univariate model and of epidemiological inter-
est were included in the multivariate model. Hazard ratios Demographic characteristics of the follow-up cohort
(HR) and 95% confidence intervals (CI) were used to assess compared to the lost-follow-up cohort
the risk of developing TB. Data analysis was performed The mean age of the study cohort was 73.0 (± 6.36 years
using SPSS software (version 17.0, SPSS for Windows Re- old, ranged from 65 to 108 years old) and about two third
lease 17.0, SPSS Inc., Chicago, Illinois, USA). were aged from 65 to 74 years. We compared the demo-
Based on the results from multivariate analysis, we calcu- graphic characteristics of the elderly cohort followed up
lated the population attributable fraction (PAF) of each risk with those lost to follow-up during the study period, given
factor contributing to incident TB among elderlies. We the high proportion of elderlies deceased, refused to
used the adjusted HR and baseline prevalence of each asso- follow-up examination and moved or transferred out
ciated factor among elderlies to estimate its contribution to (Table 1). There were significant more elderlies aged 75
bacteriologically positive and active TB, respectively. and above (40.3%, P < 0.001) lost to follow up. Of the lost

Fig. 2 Number of elderly people followed up, 2013–2015

Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 6 of 13

Table 1 Comparison of characteristics between the follow-up Table 1 Comparison of characteristics between the follow-up
and lost-to-follow up cohorts, 2013–2015* and lost-to-follow up cohorts, 2013–2015* (Continued)
Variables Total Follow-up Lost-to- P Variables Total Follow-up Lost-to- P
follow up follow up
N (%) n (%) n (%) N (%) n (%) n (%)
Total 34 076 22 119 11 957 Yes 77 (0.2) 63 (0.3) 14 (0.1)
Gender Unknown 10 (0.0) 4 (0.0) 6 (0.1)
Male 15 913 (46.7) 10 385 (47.0) 5528 (46.2) 0.207 Cigarette smoker
Female 18 163 (53.3) 11 734 (53.0) 6429 (53.8) Non-smoker 27 363 (80.3) 17 544 (79.3) 9819 (82.1) < 0.001
Age groups Ex/current 6707 (19.7) 4571 (20.7) 2136 (17.9)
smoker
65–74 21 592 (63.4) 14 452 (65.3) 7140 (59.7) < 0.001
Unknown 6 (0.0) 4(0.0) 2(0.0)
≥ 75 12 484 (36.6) 7667 (34.7) 4817 (40.3)
Alcohol drinker
Nationality
Non-drinker 27 526 (80.9) 17 604 (79.6) 9922 (83.0) < 0.001
Han 29 554 (86.7) 18 918 (85.5) 10 636 (89.0) < 0.001
Ex/current 6501 (19.1) 4499 (20.3) 2002 (16.7)
Others 4507 (13.3) 3198 (14.5) 1309 (10.9)
drinker
Unknown 15(0.1) 3 (0.0) 12 (0.1)
Unknown 49 (0.1) 16 (0.1) 33 (0.3)
Education
BMI level
Illiteracy 12 016 (35.3) 8645 (39.1) 3371 (28.2) < 0.001
18.5–23.9 20 754 (60.9) 13 680 (61.8) 7074 (59.2) < 0.001
Primary to 17 168 (50.4) 11 413 (51.6) 5755 (48.1)
< 18.5 3585 (10.5) 2477 (11.2) 1108 (9.3)
secondary
≥ 24 9730 (28.6) 5958 (26.9) 3772 (31.5)
High school 3663 (10.7) 1343 (6.1) 2320 (19.4)
and above Unknown 7 (0.0) 4 (0.0) 3 (0.0)
Unknown 1229 (3.6) 718 (3.2) 511 (4.3) # Others included single, divorced and widowed. TB Tuberculosis, BMI Body
mass index
Marital status
Married 24 828 (72.9) 15 970 (72.2) 8858 (74.1) 0.001
to follow-up cohort, most elderly (86.7%) were Han na-
Others# 9244 (27.1) 6147 (27.8) 3097 (25.9)
tionality (P < 0.001). A higher proportion of lost to follow
Unknown 4 (0.0) 2 (0.0) 2 (0.0)
up presented in elderlies with higher education (6.1% vs
Residence 19.4%; P < 0.001). Elderlies without spouse had higher
Local resident 31 316 (91.9) 21 487 (97.1) 9829 (82.2) < 0.001 proportion of lost to follow-up (P < 0.01). A greater pro-
Others 2760 (8.1) 632 (2.9) 2128 (17.8) portion of elderlies who were not local residents lost to
Annual average income per person in the family (RMB) follow up during the study period (17.8% versus 2.9%;
P < 0.001). A substantial proportion (64.9%) of elderly
≥ 10 000 16 506 (48.4) 8748 (39.5) 7758 (64.9) < 0.001
people who reported higher annual average income lost to
2300–9999 11 935 (35.0) 9230 (41.7) 2705 (22.6)
follow up (P < 0.001). About 10% of the cohort had a BMI
< 2300 4010 (11.8) 3058 (13.8) 952 (8.0) less than 18.5, and a greater proportion with BMI ≥ 24 lost
Unknown 1625 (4.8) 1083 (5.0) 543 (4.5) to follow up (P < 0.001).
Diabetes
No 31 690 (93.0) 20 697 (93.6) 10 993 (91.9) < 0.001 Incidence rate of bacteriologically positive TB and
associated risk factors among elderly
Yes 2386 (7.0) 1422 (6.4) 964 (8.1)
Over the follow-up period, the observed cumulative PY
Previously treated with TB
for the elderly cohort was 44 622.2. A total of 62 bac-
No 33 503 (98.3) 20 530 (98.4) 11 750 (98.3) 0.367 teriologically positive incident TB cases were identified
Yes 573 (1.7) 333 (1.6) 207 (1.7) and the incidence rate was 138.9 per 100 000 PY (95%
Chronic bronchitis CI: 104.4–173.5 per 100 000 PY). Among the 62 incident
No 32 346 (94.9) 20 905 (94.5) 11 441 (95.7) < 0.001 cases, 79.0% (49) were detected by active case finding,
compared to 21.0% (13) by passive case finding
Yes 1632 (4.8) 1147 (5.2) 485 (4.1)
(P < 0.001). The incidence rate for elderly with different
Unknown 98 (0.3) 67 (0.3) 31 (0.3)
characteristics and HRs were calculated and presented
Pneumoconiosis in Table 2. Those elderlies, who were male, aged older
No 33 989 (99.7) 22 052 (99.7) 11 937 (99.8) 0.002 than 75 years, Han nationality, illiteracy, married, living
outside the study areas, annual average income per
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 7 of 13

Table 2 Incidence rate of bacteriologically positive TB and associated risk factors among the elderly followed up, 2013–2015#
Variables Person-year TB cases (n) Incidence(1/100 000 person-year) HRc (95% CI) HRa (95% CI) P
(95% CI)
Total 44 622.2 62 138.9 (104.4–173.5)
Gender
Female 23 959.1 23 96.0 (56.8–135.2) 1.00 1.00
Male 20 663.1 39 188.7 (129.5–248.0) 1.97 (1.18–3.30)** 1.47 (0.80–2.69) 0.210
Age group
65–74 28 910.3 33 114.2 (75.2–153.1) 1.00 1.00
≥ 75 15 711.9 29 184.6 (117.4–251.8) 1.63 (0.99–2.69) 1.50 (0.90–2.48) 0.117
Nationality
Han 38 909.5 61 156.8 (117.4–196.1) 1.00
Others 5703.3 1 17.5 (0.4–97.7) 0.21 (0.03–1.52)
Education
High school and above 3536.9 4 113.1 (30.82–289.5) 1.00
Primary to secondary 23 111.4 30 129.8 (83.4–176.3) 0.81 (0.29–2.32)
Illiteracy 16 550.9 28 169.2 (106.5–231.8) 1.19 (0.42–3.40)
Marital status
Married 32 515.4 49 150.7 (108.5–192.9) 1.00
Others$ 12 101.9 13 107.4 (49.0–165.8) 0.77 (0.42–1.42)
Residence
Local residents 42 495.9 58 136.5 (101.4–171.6) 1.00
Others 2126.3 4 188.1 (51.3–481.6) 2.46 (0.89–6.84)
Annual average income per person in the family (RMB)
≥ 10 000 19 617.4 22 112.2 (65.3–159.0) 1.00
2300–9999 17 234.1 28 162.5 (102.3–222.6) 1.35 (0.77–2.36)
< 2300 5708.9 12 210.2 (91.3–329.1) 1.95 (0.96–3.96)*
Diabetes
No 41 653.3 62 148.9 (111.8–185.9) 1.00
Yes 2968.9 0 0.0 (0–124.29) 0.04 (0.00–3.73)
Previously treated with TB
No 43 871.8 60 136.8 (102.2–171.4) 1.00
Yes 750.4 2 266.5 (32.3–962.2) 1.70 (0.42–6.97)
Chronic bronchitis
No 42 311.8 58 137.1 (101.8–172.4) 1.00
Yes 2188.2 4 182.8 (49.8–468.0) 1.26 (0.46–3.49)
Pneumoconiosis
No 44 500.5 62 139.3 (104.6–174.0) 1.00
Yes 112.6 0 0.0 (0.0–3275.9) -#
cigarette smoker
Non-smoker 35 608.9 38 106.7 (72.8–140.7) 1.00 1.00
Ex/current smoker 9006.3 24 266.5 (159.9–373.1) 2.27 (1.36–3.78)** 2.12 (1.27–3.54) 0.004
alcohol drinker
Non-drinker 35 752.6 48 134.3 (96.3–172.2) 1.00
Ex/current drinker 8827.0 14 158.6 (75.5–241.7) 1.06(0.58–1.92)
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 8 of 13

Table 2 Incidence rate of bacteriologically positive TB and associated risk factors among the elderly followed up, 2013–2015#
(Continued)
Variables Person-year TB cases (n) Incidence(1/100 000 person-year) HRc (95% CI) HRa (95% CI) P
(95% CI)
BMI level
18.5–23.9 27 380.2 39 142.4 (97.7–187.1) 1.00 1.00
< 18.5 4789.2 17 355.0 (186.2–523.7) 2.37 (1.34–4.19)** 2.33 (1.32–4.12) 0.004
≥ 24 12 445.2 6 48.2 (17.7–104.9) 0.33 (0.14–0.77)** 0.34 (0.14–0.80) 0.014
Those with missing values were excluded from analysis; TB Tuberculosis, HRc crude hazard ratio; HRa hazard ratio adjusted; *P ≤ 0.05; ** P < 0.01; *** P < 0.001.
$
Others included single, divorced and widowed. # Unavailable HRc because of no case for this group

person in the family less than RMB 2300, previously developing active TB. And, BMI ≥ 24 decreased the risk
treated with TB, suffering chronic bronchitis, ex of developing active TB by 47%.
−/current smoker and BMI less than 18.5 had a much
higher TB incidence rates over the follow-up period. Population attributable factors to TB incidence among
In the univariate model, elderlies who were male, ex elderlies followed up
or current smoker and BMI < 18.5 had higher risk in de- We calculated the PAF of identified risk factors for TB
veloping incident TB. BMI ≥ 24 presented as a protection using adjusted HR obtained from multivariate analyses
factor for developing bacteriologically positive TB dis- (Table 4). For bacteriologically positive TB, self-reported
ease. All statistically significant variables in univariate ex or current smoker had the biggest contribution
analysis were included into multivariate analyses to cal- (18.1%) followed by male (18.0%) and age at 75 years and
culate the adjusted HR. In the multivariate model, ex or above (15.4%). For active TB, non-Han nationality
current smoker (HR = 2.12, 95% CI 1.27–3.54; P < 0.01) (35.4%) had the biggest contribution followed by male
and BMI < 18.5 (HR = 2.33, 95% CI: 1.32–4.12; P < 0.01) (26.8%) and age at 75 years and above (10.8%).
were the strong predictors in developing bacteriologic-
ally positive TB disease. For those elderlies with BMI ≥ Discussion
24.0, their risks in developing bacteriologically positive The high TB incidence obtained from our prospective
TB were substantially reduced by 66%. cohort study highlighted the importance of active case
finding among elderly in China. The results indicated
that male, non-Han nationality, previously treated with
Incidence rate of active TB and associated risk factors TB, ex/current smoker and BMI < 18.5 were risk factors
among elderly for developing TB disease among elderly in China.
During the study period, a total of 215 active TB cases TB incidence is a key indicator to evaluate TB
were detected and the incidence rate was 481.8 per 100 epidemic and the effectiveness of control strategy at
000 PY (95% CI: 417.4–546.2 per 100 000 PY) (Table 3). country level as set in the END TB strategy by WHO
Among the 215 active cases, 83.7% (180) were detected [21–23]. Although cohort study, as a kind of prospective
by active case finding, compared to 16.3% (35) by passive study, had advantages of identifying causality and yield-
case finding (P < 0.001). Those elderlies, who were male, ing incidence, this direct measurement of TB incidence
aged 75 years and older, non-Han nationality, illiteracy, is prohibitively complicated both logistically and finan-
married, local residents, annual average household cially, therefore, indirect estimation of TB incidence is
income per person less than RMB 2300, not diabetes, widely used [23]. For countries with reliable surveillance
previously treated with TB, chronic bronchitis, pneumo- system, directly notification rate is used as incidence
coniosis, ex or current smoker, ex-drinker or current rate, including reporting the epidemiological status for
drinker and BMI < 18.5 had higher incidence rates. elderly TB. In the literature we reviewed, TB incidence
In univariate analyses, male, age group at 75 years and rate among elderlies was various in countries with high
older, non-Han nationality, annual average household and low TB burden. For the United States and Germany,
income per person less than RMB 10 000 Yuan, previ- the reported average yearly rate of TB disease among
ously treated with TB, self-reported smoker and drinker aged 65 years and above was 10.9 per 100 000 [24] and
presented statistically significant risk of developing active 11.2 per 100 000 [25], respectively. In South Africa, the
TB among elderly. BMI ≥ 24 was a statistically significant reported incidence rate was 518–684 per 100 000 for
protection factor for elderlies from developing active TB. male and 193–314 per 100 000 for female among people
In the multivariate model, male, non-Han nationality, aged 65 years and older [26]. An eight-year follow-up
previously treated TB and ex or current smoker were in- study conducted among the elderlies in Taiwan Province
dependently associated with the increased risk of reported similar incidence rate (175.5 per 100 000
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 9 of 13

Table 3 Incidence rates and relative risk for active TB among elderlies followed up, 2013–2015#
Variables Person-year TB cases (N) Incidence(1/100 000 HRc (95% CI) HRa (95% CI) P
person-year (95% CI)
Total 44 622.2 215 481.8 (417.4–546.2)
Gender
Female 23 959.1 76 317.2 (245.9–388.5) 1.00 1.00
Male 20 663.1 139 672.7 (560.9–784.5) 2.11 (1.59–2.80)*** 1.78 (1.27–2.50) 0.001
Age group
65–74 28 910.3 121 418.5 (344.0–493.1) 1.00 1.00
≥ 75 15 711.9 94 598.3 (477.3–719.2) 1.42 (1.08–1.87)* 1.33 (1.00–1.80) 0.053
Nationality
Han 38 909.5 157 403.5 (340.4–466.6) 1.00 1.00
Others 5703.3 58 1017.0 (755.2–1278.7) 5.29 (3.77–7.41)*** 5.15 (3.52–7.54) < 0.001
Education
High school and above 3536.9 12 339.3 (147.3–531.3) 1.00
Primary to secondary 23 111.4 108 467.3 (379.2–555.4) 1.00 (0.55–1.82)
Illiteracy 16 550.9 95 574.0 (458.6–689.4) 1.36 (0.74–2.49)
Marital status
Married 32 515.4 168 516.7 (438.6–594.8) 1.00
Others$ 121 01.9 47 388.4 (277.3–499.4) 0.82 (0.59–1.13)
Residence
Local residents 42 495.9 207 487.1 (420.8–553.5) 1.00
Others 2126.3 8 376.2 (162.3–741.2) 1.44 (0.71–2.92)
Annual average income per person in the family (RMB)
≥10 000 19 617.4 63 321.1 (241.8–400.5) 1.00 1.00
2300–9999 17 234.1 93 539.6 (430.0–649.3) 1.53 (1.11–2.11)*** 1.32 (0.95–1.83) 0.095
< 2300 5708.9 41 718.2 (498.4–938.0) 2.29 (1.54–3.40)*** 1.48 (0.97–2.26) 0.072
Diabetes
No 41 653.3 206 494.6 (427.0–562.1) 1.00
Yes 2968.9 9 303.1 (138.8–575.3) 0.62 (0.32–1.21)
Previously treated with TB
No 43 871.8 201 458.2 (394.8–521.5) 1.00 1.00
Yes 750.4 14 1865.7 (888.4–2843.0) 3.70 (2.15–6.37)*** 3.18 (1.83–5.52) < 0.001
Chronic bronchitis
No 42 311.8 199 470.3 (405.0–535.7) 1.00
Yes 2188.2 16 731.2 (372.9–1089.5) 1.55(0.93–2.59)
Pneumoconiosis
No 44 500.5 213 478.7 (414.4–542.9) 1.00
Yes 112.6 2 1755.6 (214.8–6409.8) 3.65 (0.91–14.71)
cigarette smoker
Non-smoker 35 608.9 146 410.0 (343.5–476.5) 1.00 1.00
Ex/current smoker 9006.3 69 766.1 (585.4–946.9) 1.77 (1.33–2.37)*** 1.48 (1.06–2.09) 0.023
alcohol drinker
Non-drinker 35 752.6 156 436.3 (367.9–504.8) 1.00 1.00
Ex/current drinker 8827.0 59 668.4 (497.9–839.0) 1.37 (1.01–1.85)* 0.941 (0.66–1.34) 0.732
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 10 of 13

Table 3 Incidence rates and relative risk for active TB among elderlies followed up, 2013–2015# (Continued)
Variables Person-year TB cases (N) Incidence(1/100 000 HRc (95% CI) HRa (95% CI) P
person-year (95% CI)
BMI level
18.5–23.9 27 380.2 147 536.9 (450.1–623.7) 1.00 1.00
< 18.5 4789.2 37 772.6 (523.6–1021.5) 1.39 (0.97–2.00) 1.29 (0.88–1.88) 0.194
≥ 24 12 445.2 31 249.1 (161.4–336.8) 0.46 (0.31–0.67)*** 0.53 (0.35–0.79) 0.002
#Those with missing values were excluded from analysis; TB Tuberculosis, HRc crude hazard ratio, HRa hazard ratio adjusted; *P ≤ 0.05; ** P < 0.01; *** P < 0.001;
$
Others included single, divorced and widowed

person-year) [27] to our study. The TB incidence rate baseline, we did not perform clinical measurement.
we reported was higher than the rate reported by using Moreover, about half of known diabetes patients lost to
IDRS data (about 155/100000 for the elderlies) due to follow up during follow-up period.
the implementing of yearly active casing finding. To our knowledge, this is the first study that investi-
Elderlies are generally at risk to develop TB due to gated the contribution of risk factors to incident TB
compromised immune responses and reactivation of among elderlies in China. We obtained PAF of each risk
previous “latent TB” or new TB infection [25, 28, 29]. factor identified in our study to incident TB among eld-
Our study identified that age, male [30], smoking [31], erly. A study conducted in 22 high TB burden countries
previously treated TB, BMI < 18.5 [32] and low annual mostly in Africa and Asia reported that the top three
household income as main risk factors for elderly TB, contribution risk factors were malnutrition (27.0%),
being very similar to other studies globally [30–32]. A smoking (21.0%) and HIV infection (16.0%) [39]. In our
cohort study conducted in Taiwan Province reported study, the biggest contributor for elderly developing bac-
that age older than 70 years, male, living in rural areas, teriologically positive TB was ex−/current smoker
diabetes, congestive heart failure, chronic obstructive followed by low BMI. For developing active TB, the big-
pulmonary disease, chronic kidney disease and cancer gest contributors were non-Han nationality, male sex
were independent risk factors for incident TB among el- and smoking. This finding provide the basic data for
derlies. This also highlighted the impact of TB co- evaluating the impact of different intervention strategy,
morbidity among elderlies [33–35]. However, we did not and has an important public health implication. It will
identify diabetes as a risk factor for developing TB dis- be very useful for policy maker, especially in resource
ease among our study population, as reported by studies limited regions.
conducted in Mexico [36] and Korea [37], and also in We observed a high proportion of lost-to-follow-up
Taiwan Province of China [27]. There were several rea- over the two-year study period, inevitably producing ef-
sons underlay it. Firstly, the percentage of diabetes re- fect on results. In our study, the main reason of high
ported among our elderly cohort was 7.0%, which was lost-to-follow-up was that participants were healthy el-
much lower than reported prevalence, that is, diabetes derlies without TB, resulting in a high proportion (about
prevalence being 20.2% in Chinese people aged 60 years 30%) of them refused the follow-up check. High propor-
and above, and only 36.5% of diabetes patients being tion of lost-to-follow-up was also reported by other co-
aware of their diagnosis [38], resulting in under- hort studies, even if participants were TB patients. A
reporting of diabetes. Secondly, in our study, diabetes community-based cohort study conducted among people
status was mainly self-reported by participants at aged older than 14 years in Southern Ethiopia reported

Table 4 Contribution of risk factors for bacteriologically positive TB and active TB

Risk factors Prevalence Bacteriologically positive TB Active TB
among
HRa PAF (%) HRa PAF (%)
cohort
Male 0.467 1.470 18.0 1.784 26.8
Age(≥ 75) 0.366 1.498 15.4 1.332 10.8
Other nationality 0.132 - - 5.152 35.4
Previously treated with TB 0.017 - - 3.175 3.6
Ex/current smoking 0.197 2.119 18.1 1.484 8.7
BMI < 18.5 0.105 2.328 12.2 1.285 2.9
-: Not applicable; TB Tuberculosis, PAF Population attributable fraction, HR Hazard ratio, BMI Body mass index;
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 11 of 13

about 1.0% lost-to-follow-up during the first half year screening strategy, to reach the first milestone and the
over the follow-up period [34], and 6.9% of TB patients targets set in the End TB Strategy.
lost to follow-up during 2 years was reported by a study
conducted in Colombian prisons [40]. In older TB pa- Conclusions
tients, the proportion of death and lost-to-follow-up hit From our finding, we conclude that the elderly popula-
12.3% and transfer-out was 3.9%, indicated by a study tion in China had a high TB incidence rate and risk to
conducted in India [41]. Another reason was that about develop TB disease. Given the nature of elderly TB, ac-
8.1% of participants was internal migrants who moved tive case finding should be applied among elderlies to
around after living in a place for some time, and 77% of detect more TB cases. Our findings from China, a coun-
them lost to follow up. In addition, another 5% of our try with high TB burden also could be referred by other
study population was transferred out during the follow- countries with aging population and high TB burdens in
up period. order to achieve the targets set in the End TB Strategy.
As a well-designed study with a large number of par-
Abbreviations
ticipants in China, our study has the following
BMI: Body mass index; CI: Confidence interval; CXR: Chest X-ray; HR: Hazard
strengths. Firstly, it was designed as a prospective co- ratio; NTP: National TB Program; PAF: Population attributable fraction;
hort study that captured incident TB cases among our PY: Person-year; TB: Tuberculosis; WHO: World Health Organization
cohort and obtained the true TB incidence by perform-
Acknowledgements
ing regular CXR in the follow-up period. Secondly, our We thank the tireless contributions of the staff in the provincial CDCs, local
study was a random sampled population-based study. It CDCs, and other related health care workers for undertaking this hard
avoided the selection bias existing such as in hospital- prospective cohort study. The study sites were located in Jiangsu Province,
Zhejiang Province, Guangdong Province and Shanghai of eastern China,
based studies, which possibly resulted in a high yield of Henan Province, Heilongjiang Province and Hubei Province of central China,
incident cases as people with risk factors for TB are and Sichuan Province, Guangxi Zhuang Autonomous Region and Yunnan
likely visiting hospitals. For example, in two pilot stud- Province of western China.

ies conducted in China to screen TB among diabetes Authors’ contributions

patients, the clinic based study yielded higher TB notifi- JC was responsible for study design, implementation of this study, data
cation than the community one [42, 43]. Finally, our management and quality control, data analysis and interpretation, literature
search, figures making, initial conception and writing for original draft. YNS
study had a very strict quality control including internal was responsible study design, data interpretation, literature search, figures
quality control measurements and external supervision making and writing original draft. CYZ was responsible for implementation
for study procedure and data collection to ensure the of this study, data management and quality control, data analysis, literature
search, figures making and comments for this draft. YLY, LHT, HP, YP, YXY,
study quality. SYH, JWL, JMZ, LXia and LXu were responsible for organizing investigation at
Our study has several limitations. First, a high propor- the study sites, data collection and data management, and comments for
tion of lost-to-follow-up presented among the elderlies this draft. YYX and FZ were responsible for implementation of this study,
data management and quality control, and comments for this draft. LXW
with TB risk factors. This might result in an underesti- and HZ were responsible for funding acquisition and study design,
mation of TB incidence among elderlies. Second, due to supervision, critical revision for original draft and final decision for
the restriction of funding and resources, we could only submission. All authors contributed to review and have seen and approved
this manuscript for submission.
conduct a two-year follow-up. However, TB is a chronic
infectious disease and the onset of disease took time and Funding
only happened once immunity weakened. Two-year fol- This study was supported by The National Twelfth Five-year Mega-Scientific
Projects of infectious diseases of China (grant No.: 2013ZX10003004–001),and
low up was therefore not long enough to capture all TB
the funder had no contribution to study design, data collection and analysis,
incidences. We recommend longer follow up among the result interpretation and paper writing.
elderlies in future studies.
Although with some limitations, in the context of end- Availability of data and materials
The National Center for Tuberculosis Control and Prevention (NCTB) is the
ing TB, our population-based prospective cohort study custodian of the data for this study. The data are not accessible online, but
reported a very important area in TB control in a coun- may be made available upon written request to the NCTB through the
try with high TB burden. By investigating the TB inci- authors, if in line with the Ethical Review Board guidelines.
dence among elderly in China, we provided scientific Ethics approval and consent to participate
evidence for the urgent need in conducting active case The study protocol was approved by the Research Ethics Review Committee
finding among elderly, a group of TB patients who often of Chinese Center for Disease Control and Prevention (Approval number:
201322). The signed informed consent was obtained from all participants
do not present TB symptoms. The PAF for TB identified before investigation.
the key population to implement intervention strategies
in order to rapidly decrease the TB incidence rate. Based Consent for publication
Not applicable.
on our study, we could further observe the long term
impact of active case finding on local TB epidemic Competing interests
among elderlies and further provide the optimal The authors declare that they have no competing interests.
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 12 of 13

Author details 16. Cheng J, Wang L, Zhang H, Xia Y. Diagnostic value of symptom screening
1
National Center for Tuberculosis Control and Prevention, Chinese Center for for pulmonary tuberculosis in China. PLoS One. 2015;10(5):e0127725. https://
Disease Control and Prevention, Beijing, China. 2National Center for doi.org/10.1371/journal.pone.0127725.
Population Health and Epidemiology, Canberra, Australia. 3Heilongjiang 17. Zhang C, Zhao F, Xia Y, Yu Y, Shen X, Lu W, et al. Prevalence and risk factors
Provincial Center for Disease Control and Prevention, Harbin, China. of active pulmonary tuberculosis among elderly people in China: a
4
Minhang District Center for Disease Control and Prevention, Shanghai, population based cross-sectional study. Infect Dis Poverty. 2019;8(7). https://
China. 5Jiangsu Provincial Center for Disease Control and Prevention, doi.org/10.1186/s40249-019-0515-y.
Nanjing, China. 6Zhejiang Provincial Center for Disease Control and 18. Chen W, Shu W, Wang M, Hou Y, Xia Y, Xu W, et al. Pulmonary tuberculosis
prevention, Hangzhou, China. 7Henan Provincial Center for Disease control incidence and risk factors in rural areas of china: a cohort study. PLoS ONE.
and prevention, Zhengzhou, China. 8Hubei Provincial Center for Disease 2013;8(3):e58171. https://doi.org/10.1371/journal.pone.0058171.
Control and Prevention, Wuhan, China. 9Center for Tuberculosis Control of 19. Xiao D, Zhao M, Wang Y, Wang L, Xu S, Wang W. Guidelines for
Guangdong Province, Guangzhou, China. 10GuangXi Center for Disease implementing the national tuberculosis control program in China (2008).
Prevention and Control, Nanning, China. 11Sichuan Provincial Center for Beijing: Union Medical College Press; 2009.
Disease Control and Prevention, Chengdu, China. 12Yunnan Provincial Center 20. National Health Commission of the People’s Republic of China. Criteria of
for Disease Control and Prevention, Kunming, China. weight for adults: National Health Commission of the People’s Republic of
China. Available from: http://www.nhc.gov.cn/ewebeditor/uploadfile/2013/
Received: 13 June 2019 Accepted: 9 December 2019 08/20130808135715967.pdf. Accessed 8 Aug 2019.
21. World Health Organization. The End TB Strategy. Available from: https://
www.who.int/tb/End_TB_brochure.pdf?ua=1. Accessed 8 Aug 2019.
22. Zhu S, Xia L, Yu S, Chen S, Zhang J. The burden and challenges of
References tuberculosis in China: findings from the global burden of disease study
1. World Health Organization. World report on aging and health. Available 2015. Sci Rep. 2017;7:14601. https://doi.org/10.1038/s41598-017-15024-1.
from: https://apps.who.int/iris/bitstream/handle/10665/186463/9789240694 23. Avilov KK, Romanyukha AA, Borisov SE, Belilovsky EM, Nechaeva OB, Karkach
811_eng.pdf?sequence=1. Accessed 8 Aug 2019. AS. An approach to estimating tuberculosis incidence and case detection
2. World Health Organization. WHO report 2007 Global tuberculosis control: rate from routine notification data. Int J Tuberc Lung Dis. 2015;19(3):288–94.
surveillance, planning, financing. Available from: https://apps.who.int/iris/ https://doi.org/10.5588/ijtld.14.0317.
bitstream/handle/10665/43629/9789241563141_eng.pdf?sequence=1. 24. Hochberg NS, Horsburgh CR. Prevention of tuberculosis in older adults in the
Accessed 8 Aug 2019. United States: obstacles and opportunities. Clin Infect Dis. 2013;56:1240–7.
3. World Health Organization. Global tuberculosis report 2017. Available from: https://doi.org/10.1093/cid/cit027.
https://www.who.int/tb/publications/2017/en/. Accessed 8 Aug 2019. 25. Hauer B, Brodhun B, Altmann D, Fiebig L, Loddenkemper R, Haas W.
4. Davies PD. TB in the elderly in industrialised countries. Int J Tuberc Lung Tuberculosis in the elderly in Germany. Eur Respir J. 2011;38(2):467–70.
Dis. 2007;11(11):1157–9. https://doi.org/10.1183/09031936.00199910. https://doi.org/10.1183/09031936.00199910.
5. Wang L, Zhang H, Ruan Y, Chin DP, Xia Y, Cheng S, et al. Tuberculosis 26. Nanoo A, Izu A, Ismail NA, Ihekweazu C, Abubakar I, Mametja D, et al.
prevalence in China, 1990-2010: a longitudinal analysis of national Nationwide and regional incidence of microbiologically confirmed
survey data. Lancet. 2014;383:2057–64. https://doi.org/10.1016/S0140- pulmonary tuberculosis in South Africa, 2004-12: a time series analysis.
6736(13)62639-2. Lancet Infect Dis. 2015;15:1066–76. https://doi.org/10.1016/S1473-
6. World Health Organization. China country assessment report on ageing and 3099(15)00147-4.
health. Available from: https://apps.who.int/iris/bitstream/handle/10665/1 27. Yen YF, Pan SW, Su VYF, Chuang PH, FengJY SWJ. Influenza vaccination and
94271/9789241509312_eng.pdf?sequence=1. Accessed 8 Aug 2019. incident tuberculosis among elderly persons, Taiwan. Emerg Infect Dis.
7. Pratt RH, Winston CA, Kammerer JS, Kammerer JS, Armstrong LR. 2018;24(3):498–505. https://doi.org/10.3201/eid2403.152071.
Tuberculosis in older adults in the United States, 1993-2008. J Am Geriatr 28. Negin J, Abimbola S, Marais BJ. Tuberculosis among older adults - time to take
Soc. 2011;59:851–7. https://doi.org/10.1111/j.1532-5415.2011.03369.x. notice. Int J Infect Dis. 2015;32:135–7. https://doi.org/10.1016/j.ijid.2014.11.018.
8. Technical Guidance Group of the Fifth National TB Epidemiological 29. Marengoni A, Angleman S, Melis R, Mangialasche F, Karp A, Garmen A, et al.
Survey, the Office of the Fifth National TB Epidemiological Survey. The Aging with multimorbidity: a systematic review of the literature. Ageing Res
fifth national tuberculosis epidemiological survey in 2010. Chin J Rev. 2011;10:430–9. https://doi.org/10.1016/j.arr.2011.03.003.
Antituberc. 2012;34(8):485–508. 30. Leung CC, Yew WW, Chan CK, Chau CH, Tam CM, Lam CW, et al. Tuberculosis
9. Huynh GH, Klein DJ, Chin DP, Wagner BG, Eckhoff PA, Liu R, et al. in older people: a retrospective and comparative study from Hong Kong. J Am
Tuberculosis control strategies to reach the 2035 global targets in China: Geriatr Soc. 2002;50:1219–26. https://doi.org/10.1046/j.1532-5415.2002.50308.x.
the role of changing demographics and reactivation disease. BMC Med. 31. Leung CC, Li T, Lam TH, Yew WW, Law WS, Tam CM, et al. Smoking and
2015;13:88. https://doi.org/10.1186/s12916-015-0341-4. tuberculosis among the elderly in Hong Kong. Am J Respir Crit Care Med.
10. Mori T, Leung CC. Tuberculosis in the global aging population. Infect Dis 2004;170:1027–33. https://doi.org/10.1164/rccm.200404-512OC.
Clin N Am. 2010;24:751–68. https://doi.org/10.1016/j.idc.2010.04.011. 32. Jeon CY, Murray MB. Diabetes mellitus increases the risk of active
11. Vynnycky E, Borgdorff MW, Leung CC, Fine PEM. Limited impact of tuberculosis: a systematic review of 13 observational studies. PLoS Med.
tuberculosis control in Hong Kong: attributable to high risks of reactivation 2008;5:e152. https://doi.org/10.1371/journal.pmed.0050152.
disease. Epidemiol Infect. 2008;136:943–52. https://doi.org/10.1017/ 33. Siroka A, Ponce NA, Lonnroth K. Association between spending on social
S0950268807008552. protection and tuberculosis burden: a global analysis. Lancet Infect Dis.
12. Wang L, Liu X, Huang F, Hennig C, Uplekar M, Jiang S. Engaging hospitals 2016;16(4):473–9. https://doi.org/10.1016/S1473-3099(15)00401-6.
to meet tuberculosis control targets in China: using the internet as a tool to 34. Woldesemayat EM, Datiko DG, Lindtjørn B. Follow-up of chronic coughers
put policy into practice. Bull World Health Organ. 2010;88:937–42. https:// improves tuberculosis case finding: results from a community-based cohort
doi.org/10.2471/BLT.09.071753. study in southern Ethiopia. PLoS One. 2015;10:e0116324. https://doi.org/10.
13. Huang F, Cheng S, Du X, Chen W, Scano F, Falzon D, et al. Electronic 1371/journal.pone.0116324.
recording and reporting system for tuberculosis in China: experience and 35. Lönnroth K, Williams BG, Cegielski P, Dye C. A consistent log-linear
opportunities. J Am Med Inform Assoc. 2014;21:938–41. https://doi.org/10. relationship between tuberculosis incidence and body mass index. Int J
1136/amiajnl-2013-002001. Epidemiol. 2010;39:149–55. https://doi.org/10.1093/ije/dyp308.
14. Li T, Shewade HD, Soe KT, Rainey JJ, Zhang H, Du X, et al. Under-reporting 36. Ponce-De-Leon A, Garcia-Garcia MDL, Garcia-Sancho MC, Gomez Perez FJ,
of diagnosed tuberculosis to the national surveillance system in China: an Valdespino Gomez JL, Olaiz Fernandez G, et al. Tuberculosis and diabetes in
inventory study in nine counties in 2015. BMJ Open. 2019;9:e021529. southern Mexico. Diabetes Care. 2004;27:1584–90. https://doi.org/10.2337/
https://doi.org/10.1136/bmjopen-2018-021529. diacare.27.7.1584.
15. Tao L, Chen P, Chen W, Xia Y, Chen H, Zhang H, et al. Assess the sensitivity 37. Kim SJ, Hong YP, Lew WJ, Yang SC, Lee EG. Incidence of pulmonary
of China’s TB surveillance system through inventory study and capture- tuberculosis among diabetics. Tuber Lung Dis. 1995;76:529–33. https://doi.
remark-recapture method. Chin J Antitubec. 2016;38(12):1026–31. org/10.1016/0962-8479(95)90529-4.
Cheng et al. Infectious Diseases of Poverty (2020) 9:13 Page 13 of 13

38. Wang L, Gao P, Zhang M, Huang Z, Zhang D, Deng Q, et al. Prevalence and
ethnic pattern of diabetes and prediabetes in China in 2013. JAMA. 2017;
317(24):2515–23. https://doi.org/10.1001/jama.2017.7596.
39. Lönnroth K, Castro KG, Chakaya JM, Chauhan LS, Floyd K, Glaziou P, et al.
Tuberculosis control and elimination 2010-50: cure, care, and social
development. Lancet. 2010;375:1814–29. https://doi.org/10.1016/S0140-
6736(10)60483-7.
40. Rueda ZV, López L, Vélez AL, Marín D, Giraldo MR, Pulido H, et al. High
incidence of tuberculosis, low sensitivity of current diagnostic scheme and
prolonged culture positivity in fore Colombian prisons, a cohort study. PLoS
One. 2013;8(11):e80592. https://doi.org/10.1371/journal.pone.0080592.
41. Oshi D, Oshi SN, Alobu I, Ukwaja KN. Profile and treatment outcomes of
tuberculosis in the elderly in southeastern Nigeria, 2011–2012(R20). PLoS
One. 2014;9(11):e111910. https://doi.org/10.1371/journal.pone.0111910.
42. Lin Y, Innes A, Xu L, Li L, Chen J, Hou J, et al. Screening of patients with
diabetes mellitus for tuberculosis in community health settings in China.
Trop Med Int Heal. 2015;20(8):1073–80. https://doi.org/10.1111/tmi.12519.
43. Lin Y, Li L, Mi F, Du J, Dong Y, Li Z, et al. Screening patients with diabetes
mellitus for tuberculosis in China. Tropical Med Int Health. 2012;17(10):1302–8.
https://doi.org/10.1111/j.1365-3156.2012.03069.x.