ORIGINAL ARTICLE

Tuberculosis increases the risk of peripheral arterial disease:

A nationwide population-based study
SHENG-HUEI WANG,1 WU-CHIEN CHIEN,2,3 CHI-HSIANG CHUNG,2 FU-HUANG LIN,2 CHUNG-KAN PENG,1,4
CHIH-FENG CHIAN1 AND CHIH-HAO SHEN1

1
Division of Pulmonary and Critical Care Medicine, Department of Internal Medicine, Tri-Service General Hospital, 2School of
Public Health, 3Department of Medical Research, Tri-Service General Hospital and 4Institute of Undersea and Hyperbaric
Medicine, National Defense Medical Center, Taipei, Taiwan

ABSTRACT
SUMMARY AT A GLANCE
Background and objective: According to several stud-
ies, tuberculosis (TB) may be involved in the pathogen- This study showed that patients with tuberculosis
esis of cardiovascular disease. However, the relationship (TB) have a significantly higher risk of developing
between TB and peripheral arterial disease (PAD) has peripheral arterial disease (PAD) than healthy con-
not been studied. The aim of this study was to investi- trol subjects. TB should be considered when evalu-
gate whether patients with TB exhibit an increased risk ating a patient’s risk of developing PAD.
of developing PAD.
Methods: The data assessed in this national
population-based cohort study were obtained from the Abbreviations: aHR, adjusted HR; CKD, chronic kidney
Taiwan National Health Insurance Database from 2000 disease; CVD, cardiovascular disease; DM, diabetes mellitus;
to 2010. Patients with newly diagnosed TB were selected HCV, hepatitis C virus; HF, heart failure; HR, hazard ratio; ICD-9-
using the International Classification of Diseases, Ninth CM, International Classification of Diseases, Ninth Revision,
Revision, Clinical Modification (ICD-9-CM) codes. The Clinical Modification; IFN-γ, interferon-γ; IHD, ischaemic heart
non-TB cohort was randomly frequency-matched to the disease; IR, incidence density rate; IRR, incidence rate ratio;
TB cohort at a ratio of 2:1 according to age, sex and LTBI, latent TB infection; mHSP65, mycobacterial heat-shock
index year. Cox’s proportional hazards regression mod- protein 65; NHI, National Health Insurance; NHIRD, National
els were used to analyse the risk of PAD. Health Insurance Research Database; PAD, peripheral arterial
disease; TB, tuberculosis; TH, T helper.
Results: We enrolled 14 350 patients with TB and
28 700 controls in this study. The risk of PAD was 3.93-
fold higher in the patients with TB than in the non-TB INTRODUCTION
controls after adjusting for age, sex, co-morbidities and
socio-economic status. Based on the subgroup analysis, Tuberculosis (TB) was one of the top 10 causes of death
the TB cohort exhibited an increased risk of developing in the world in 2015, and the TB epidemic has been
PAD compared with the non-TB cohort, regardless of underestimated in the past few years.1 According to the
age, sex, co-morbidities and socio-economic status. World Health Organization, approximately 10.4 million
Patients with TB had a higher risk of developing PAD people developed TB and 1.4 million people died from
than healthy control subjects after 1 year of follow-up. TB worldwide in 2015.1 In Taiwan, TB is a common dis-
Conclusion: Patients with TB have a significantly higher ease, with an incidence of 53 patients per 100 000 popu-
risk of developing PAD than patients without TB. TB
lation, and 626 TB-related deaths occurred in 2012.2 TB
should be considered when evaluating a patient’s risk of
is a chronic inflammatory disease caused by Mycobacte-
developing PAD.
rium tuberculosis, which utilizes several methods to sur-
Key words: tuberculosis, peripheral arterial disease, athero- vive the intracellular environment and is involved in the
sclerosis, risk factor, inflammation. production of multiple cytokines.3–6 The diseases that
predispose individuals to TB infection include human
immunodeficiency virus (HIV) infection, diabetes,
immune-mediated inflammatory disorders, end-stage
renal disease and silicosis, among others.7–9 On the other
Correspondence: Chih-Hao Shen, Division of Pulmonary and hand, TB primarily results in pulmonary TB, which could
Critical Care Medicine, Department of Internal Medicine, Tri- lead to conditions such as chronic obstructive pulmonary
Service General Hospital, National Defense Medical Center, disease (COPD), bronchiectasis, restrictive lung diseases,
No. 325, Section 2, Cheng-Gong Road, Neihu 114, Taipei 11490,
destructive lung diseases and bronchial stenosis, and
Taiwan. Email: potato652@yahoo.com.tw
Received 1 March 2017; invited to revise 8 April 2017; revised could cause extrapulmonary TB in the lymph nodes,
14 May 2017; accepted 19 May 2017 (Associate Editor: Chi Chiu pleural cavity, bones or joints, genitourinary tract,
Leung). meninges or peritoneum (infection sites listed in order of

© 2017 Asian Pacific Society of Respirology Respirology (2017) 22, 1670–1676

doi: 10.1111/resp.13117
TB increases the risk of PAD 1671

proportion).10 Furthermore, some systemic diseases that Information). We defined the index date as the date
seem to be related to TB have been reported, such as of TB diagnosis. The follow-up period was defined as
acute coronary syndrome, liver cirrhosis, systemic lupus the time from the index date to the date of PAD diag-
erythematosus and Parkinson’s disease.11–14 nosis. The exclusion criteria included a history of TB
Peripheral arterial disease (PAD) is a disease of the or PAD before the index date, age < 20 years and
circulatory system that decreases blood flow to periph- missing information (age or sex). The non-TB cohort
eral tissues, is primarily caused by atherosclerosis and was randomly selected from the NHIRD using the
induces tissue injury. The prevalence of PAD is approx- same exclusion criteria. Two patients from the non-
imately 12–14% in the general population, and the TB cohort were matched with each patient included
prevalence increases with age to affect 20% of patients in the TB cohort according to sex, age and
over 75 years.15 In 2010, PAD affected 202 million peo- index year.
ple globally, and it currently remains one of the most
critical public health issues.16 Common risk factors for
PAD include old age, hypertension, type 2 diabetes,
smoking, hypercholesterolaemia and obesity.17,18 Some End point and co-morbidities
infectious diseases, including pneumonia, chronic The end point of this study was the development of a
bronchitis, chickenpox, shingles and mumps, have also
new diagnosis of PAD (ICD-9-CM 440.0, 440.2–440.3,
been reported to be independently associated with
440.8–440.9, 443, 444.0, 444.22, 444.8 and 447.8–447.9)
PAD.19,20
since the index date (Table S2, Supplementary Infor-
Inflammation participates in all stages of atheroscle-
mation). The diagnosis of PAD was based on the ICD-
rosis, from initiation and progression to complications, 9-CM and was determined by relevant specialists
and leads to the development of various circulatory
according to typical symptoms, laboratory data and
diseases, including PAD.21–24 TB infection is an inflam-
imaging of PAD. All patients were followed up from the
matory process involving interactions between the
index date to the diagnosis of PAD, withdrawal from
inflammatory response and suppressive mediators.25,26 NHI programme, death or 31 December 2010. The co-
Furthermore, the role of TB in the development of car-
morbidities we used included diabetes mellitus (DM;
diovascular disease (CVD) epidemics and pathogenesis
ICD-9-CM 250), hypertension (ICD-9-CM 401–405),
has been discussed in recent years.27 However, the
hyperlipidaemia (ICD-9-CM 272), CVD, stroke (ICD-9-
relationship between TB and PAD has not been stud-
CM 430–438), COPD (ICD-9-CM 490–491, 495–496),
ied. We conducted a longitudinal nationwide asthma (ICD-9-CM 493), chronic kidney disease (CKD;
population-based cohort study to investigate whether
ICD-9-CM 585), HIV (ICD-9-CM 042, 043, 044 and
TB increased the risk of PAD.
V08) infection, hepatitis C virus (HCV; ICD-9-CM
070.41, 070.44, 070.51, 070.54, 070.7 and V02.62) infec-
METHODS tion, urbanization level and monthly income (insured
premium). CVD included ischaemic heart disease
(IHD; ICD-9-CM 401–414), heart failure (HF; ICD-9-
Data source CM 428) and atrial fibrillation (AF; ICD-9-CM 427.31)
This study was conducted using data from the Longitu- to address the collinear nature of these diseases.
dinal Health Insurance Database (LHID), which was Because smoking is a common risk factor for PAD that
derived from Taiwan’s National Health Insurance was not identified in the NHIRD, we selected smoking-
(NHI) programme. This programme has been imple- related diseases, including IHD, stroke, COPD and
mented by the government since 1995 and provides asthma for adjustments to minimize the confounding
health insurance to approximately 99% of the 23.74 effect of smoking.30,31 Patients were considered to have
million Taiwanese residents. The identification of dis- a co-morbidity if they were diagnosed before the
eases in the National Health Insurance Research Data- index date.
base (NHIRD) was conducted according to the
International Classification of Diseases, Ninth Revision,
Clinical Modification (ICD-9-CM). For privacy protec-
tion, all information obtained from patients was anon-
ymized and could not be identified. All insurance Statistical analysis
claims were scrutinized by medical reimbursement We used a chi-square test to determine differences in
specialists, and the representativeness of the NHIRD demographic data and co-morbidities between the TB
for the entire Taiwanese population has been validated and non-TB cohorts. Cumulative incidence curves of
by several studies.28,29 This study was approved by the PAD were estimated for both cohorts using a Kaplan–
Institutional Review Board of Tri-Service General Meier analysis, and a log-rank test was used to deter-
Hospital. mine differences between cohorts. The incidence rate
ratio of PAD in both cohorts was estimated using a
Poisson regression analysis. The adjusted hazard ratio
Study participants (HR) and 95% CI of PAD development in both cohorts
This population-based retrospective cohort study were estimated using univariate and multivariate Cox
included a TB cohort and a non-TB cohort. The TB proportional hazards regression analyses. SAS 9.4 soft-
cohort comprised patients who were newly diagnosed ware (SPSS Inc., Chicago, IL, USA) was used for the sta-
with TB (ICD-9-CM 010–018) between 1 January 2000 tistical analyses. A two-tailed P < 0.05 was considered
and 31 December 2010 (Table S1, Supplementary statistically significant.

Respirology (2017) 22, 1670–1676 © 2017 Asian Pacific Society of Respirology

1672 S-H Wang et al.

RESULTS HR = 2.36, 95% CI = 1.01–16.44, respectively). As

shown in Figure 1, patients in the TB cohort had a sig-
We enrolled 14 350 patients in the TB cohort and nificantly higher cumulative incidence of PAD than
28 700 patients in the non-TB cohort after frequency patients in the non-TB cohort during the 11-year
matching for sex, age and index year. The distributions follow-up period (log-rank test < 0.001).
of sex and age were similar in both groups. The mean The incidence density rates (IR) and adjusted HR of
follow-up period was 5.62 (SD = 3.78) years in the TB PAD based on TB status stratified by demographic fac-
cohort and 6.01 (SD = 3.85) years in the non-TB tors and co-morbidities are shown in Table 2. Patients
cohort. Compared with the non-TB cohort, patients in in the TB cohort had higher IR than patients in the
the TB cohort had a higher prevalence of DM, COPD, non-TB cohort (5.37 vs 3.73 per 1000 person-years).
HIV and HCV and a lower prevalence of hypertension, The IR increased with age in both cohorts (IR in age
hyperlipidaemia, CVD, stroke, CKD and urbanization groups 20–44, 45–69 and ≥70 years = 3.04, 5.49 and
level (Table 1). The patient selection procedure is 5.69 in the TB cohort and 1.25, 1.67 and 5.08 in the
depicted in detail in Figure S1 (Supplementary non-TB cohort, respectively). In the co-morbidity-
Information). specific analysis, patients in the TB cohort had a signif-
Patients in the TB cohort had a higher risk of devel- icantly higher HR for PAD than patients in the non-TB
oping PAD than the non-TB cohort according to the cohort, regardless of the presence of co-morbidities,
Cox regression model (adjusted HR = 3.93, 95% with the exception of the subgroups with asthma, CKD,
CI = 3.03–4.95) (Table S3, Supplementary Information). HIV, level 3 urbanization and monthly income ≥35 000
Patients with DM, CVD, stroke or HIV exhibited a sig- (NT$ (1 US$ = 31.5 NT$ in 2016)).
nificantly higher risk of developing PAD than patients The TB cohort was divided into pulmonary TB, mili-
without these co-morbidities (adjusted HR = 1.77, 95% ary TB and extrapulmonary TB according to ICD-9-CM
CI = 1.32–2.22; adjusted HR = 1.79, 95% CI = 1.32–2.18; codes, and the numbers of patients in these subgroups
adjusted HR = 1.39, 95% CI = 1.00–1.94; and adjusted were 11 954 (83.3%), 229 (1.60%) and 2519 (17.55%),

Table 1 Comparison of the demographics, characteristics and co-morbidities of the TB and non-TB cohorts

TB

No (n = 28 700) Yes (n = 14 350)

n % n % P-value

Sex 0.99
Female 8290 28.89 4145 28.89
Male 20 410 71.11 10 205 71.11
Age (years) 0.99
20–44 4996 17.41 2498 17.41
45–69 10 394 36.22 5197 36.22
≥70 13 310 46.38 6655 46.38
Co-morbidity
DM 3683 12.83 2696 18.79 <0.001
Hypertension 5305 18.48 2307 16.03 <0.001
Hyperlipidaemia 657 2.29 154 1.07 <0.001
CVD 3797 13.23 1300 9.06 <0.001
Stroke 2707 9.43 792 5.52 <0.001
COPD 2652 9.24 2455 17.11 <0.0001
Asthma 598 2.08 316 2.20 0.22
CKD 630 2.20 245 1.71 <0.001
HIV 4 0.01 84 0.59 <0.001
HCV 288 1.00 285 1.99 <0.001
Urbanization level <0.001
1 (highest) 9072 31.61 4133 28.80
2 12 850 44.77 6288 43.82
3 2223 7.75 1178 8.21
4 (lowest) 4555 15.87 2751 19.17
Insured premium (NT$) 0.07
<18 000 28 320 98.68 14 139 98.53
18 000–34 999 300 1.05 180 1.25
≥35 000 80 0.28 31 0.22

P-value (categorical variables: chi-square/Fisher’s exact test).

1 US$, 31.5 NT$ in 2016; CKD, chronic kidney disease; CVD, cardiovascular disease; DM, diabetes mellitus; HCV, hepatitis C virus;
TB, tuberculosis.

© 2017 Asian Pacific Society of Respirology Respirology (2017) 22, 1670–1676

TB increases the risk of PAD 1673

pathogenesis of both TB and atherosclerosis, which

0.08
feature similar pro-inflammatory cytokines (interleu-
kin-1 (IL-1), IL-2, IL-6, interferon-γ (IFN-γ) and tumour
Cummulative incidence of peripheral arterial disease

necrosis factor-α (TNF-α)) and immune cells (mono-

cytes, macrophages, CD4+ T helper 1 (TH1) and TH17
0.06 cells).27 Van Eeden et al. also proposed plausible mech-
anistic pathways from lung inflammation induced by
smoking, air pollution particles or pathogens to athero-
sclerosis mediated by an inflammatory cytokine
storm.35 Second, an association between increasing
0.04
titres of mycobacterial heat-shock protein 65 (mHSP65)
antibodies and elevated coronary calcification levels
was reported in humans.36 Through the anti-mHSP65
antibody, TB may exert a pathogenic effect on athero-
0.02 sclerosis mediated by molecular mimicry and autoim-
munity.27,37 Furthermore, a hypercoagulable state
induced by increased plasma fibrinogen levels,
decreased antithrombin III levels and increased platelet
aggregation has been observed in patients with TB.38
0.00 Although patients in the TB and non-TB cohorts had
different prevalences of co-morbidities, including DM,
0 1 2 3 4 5 6 7 8 9 10 11 hypertension, hyperlipidaemia, CVD, stroke, COPD,
Yaers of Follow-up CKD, HIV, HCV and urbanization level, on the index
day, TB remained an independent risk factor for PAD
Figure 1 Kaplan–Meier analysis of the cumulative incidence of in the adjusted Cox regression analysis. Moreover, male
peripheral arterial disease (PAD) between the tuberculosis (TB)
gender, old age, diabetes, CVD, stroke and HIV were
(solid line) and non-TB (dashed line) cohorts (log-rank
P < 0.001).
independent risk factors for PAD, consistent with the
findings of other studies.15,19 Patients in the TB cohort
had a higher incidence of PAD than patients in the
respectively. Table 3 shows the incidence and adjusted non-TB cohort in most of the subgroup analyses of sex,
HR of PAD for patients with different forms of age, co-morbidity and socio-economic status.
TB. Patients with pulmonary TB and extrapulmonary Miliary TB is an infectious disease caused by haema-
TB had a 3.90- and 2.11-fold higher risk of developing togenous dissemination of M. tuberculosis to vessels
PAD than patients in the non-TB cohort. throughout the body, and the effects of atherosclerosis-
Table 4 shows the incidence and adjusted HR of induced TB may theoretically have a greater influence
PAD for these two cohorts according to the follow-up than pulmonary or extrapulmonary TB.39,40 Further-
periods. The TB cohort had a 3.45-fold higher risk of more, patients with miliary TB and PAD have the same
developing PAD than the non-TB cohort for a follow- predisposing disease factors, including DM, alcoholism,
up duration of 1–5 years (adjusted HR = 3.45, 95% CKD and HIV infection.41–44 Patients in the subgroup
CI = 2.00–4.62). For a follow-up period of longer than with miliary TB had a higher IR and adjusted HR, but
5 years, the TB cohort also had a 2.86-fold higher risk this relationship was not statistically significant.
of developing PAD than the non-TB cohort (adjusted Patients with miliary TB may have shown a higher risk
HR = 2.86, 95% CI = 1.04–3.98). of developing PAD than the control group, but this dif-
ference did not reach statistical significance because
few miliary patients were enrolled in this study.
DISCUSSION It is reasonable to suppose that the diagnosis of PAD
increased when TB patients sought medical care. How-
To the best of our knowledge, this population-based ever, we analysed the cohorts stratified by follow-up
cohort study is the first to investigate the relationship year and found that the adjusted HR of PAD signifi-
between TB and PAD. In the Cox proportional hazards cantly increased during the 1–5 years and >5 years of
model, patients with TB had a 3.93-fold higher risk of the follow-up periods, but not during the first year. The
developing PAD than patients without TB after adjust- slow and chronic process by which atherosclerosis
ing for sex, age, socio-economic status and co- leads to the development of vascular disease involves
morbidities that were considered risk factors for PAD. intimal thickening followed by fibrocalcific plaque for-
The development of atherosclerosis is the major mation.45 Because most patients with TB complete
aetiology of PAD, and several infectious pathogens treatment within 1 year, we propose that the effects of
have been shown to play a role in the pathogenesis of TB on the development of PAD may involve the forma-
atherosclerosis, including HCV, Chlamydia pneumo- tion of atherosclerotic plaques on vessels affected by
niae and HIV.32–34 Based on our study, M. tuberculosis chronic inflammation and autoimmunity.
may be another infectious agent that participates in the Furthermore, in the majority of patients with active
development of atherosclerosis and, subsequently, TB, the disease occurred due to the reactivation of
PAD. We propose some mechanisms that may underlie latent TB infection (LTBI). Chronic inflammation invol-
the increased risk of PAD in patients with TB. First, ving CD4+ and CD8+ T cells, TH17 cells and cytokines
chronic inflammation plays a crucial role in the such as TNF-α and IFN-γ develops when the immune
Respirology (2017) 22, 1670–1676 © 2017 Asian Pacific Society of Respirology
1674 S-H Wang et al.

Table 2 Comparison of the incidence and aHR for PAD in the TB and non-TB cohorts stratified by sex age and co-
morbidities

TB

No Yes Compared with the non-TB cohort

Variables Event Rate Event Rate IRR (95% CI) aHR† (95% CI)

Overall 192 3.73 126 5.37 1.44 (1.38–1.48)*** 3.93 (3.03–4.95)***

Sex
Female 48 3.25 23 3.46 1.06 (1.01–1.17)*** 3.29 (1.93–5.51)***
Male 144 3.92 103 6.12 1.56 (1.51–1.60)*** 4.01 (3.08–5.22)***
Age (years)
20–44 6 1.25 7 3.04 2.43 (1.16–2.55)** 4.49 (1.32–15.09)**
45–69 25 1.67 42 5.49 3.29 (2.78–3.38)*** 8.42 (4.99–13.26)***
≥70 161 5.08 77 5.69 1.12 (1.03–1.19)*** 3.08 (2.05–4.88)***
Co-morbidity
No 150 6.64 86 8.96 1.35 (1.27–1.41)*** 3.58 (2.04–6.21)***
Yes 42 1.46 40 2.88 1.98 (1.96–1.99)*** 5.22 (1.83–7.60)***
DM
No 124 3.96 94 5.11 1.29 (1.25–1.32)*** 3.12 (2.07–4.41)***
Yes 68 3.38 32 6.29 1.86 (1.60–2.21)*** 4.21 (3.12–6.84)**
Hypertension
No 136 3.57 96 5.15 1.44 (1.39–1.48)*** 2.19 (1.31–3.86)**
Yes 56 4.19 30 6.19 1.48 (1.29–1.59)*** 4.09 (2.22–4.74)***
Hyperlipidaemia
No 184 3.69 121 5.27 1.43 (1.37–1.47)*** 2.85 (1.99–3.64)***
Yes 8 5.14 5 10.07 1.96 (1.02–2.17)** 5.21 (1.22–10.04)*
CVD
No 129 3.07 97 4.88 1.58 (1.44–1.76)*** 4.35 (1.86–7.99)***
Yes 63 6.65 29 8.04 1.21 (1.02–1.48)*** 2.72 (1.95–3.84)***
Stroke
No 160 3.48 113 5.25 1.51 (1.46–1.60)*** 3.99 (2.17–5.33)**
Yes 32 5.92 13 6.68 1.13 (1.01–1.29)*** 1.58 (1.02–3.99)*
COPD
No 177 3.91 105 5.70 1.46 (1.38–1.51)*** 3.14 (2.28–4.35)***
Yes 15 2.45 21 4.17 1.70 (1.61–2.00)*** 4.92 (2.30–7.99)***
Asthma
No 188 3.74 126 5.52 1.48 (1.42–1.52)*** 4.05 (2.17–7.55)***
Yes 4 3.49 0 0 0 0
CKD
No 185 3.92 124 5.41 1.38 (1.23–1.50)*** 2.88 (1.54–3.90)***
Yes 7 2.19 2 3.49 1.60 (0.55–2.19) 4.44 (0.27–9.78)
HIV
No 192 3.74 115 5.35 1.43 (1.02–1.98)** 3.90 (2.11–5.15)**
Yes 0 0 11 7.74 — —
HCV
No 187 3.77 119 5.41 1.44 (1.10–1.80)** 3.91 (2.10–4.97)**
Yes 5 1.29 7 2.37 1.84 (1.03–2.96)** 4.18 (1.09–6.94)*
Urbanization level
1 (highest) 56 3.86 32 5.03 1.30 (1.15–1.39)*** 2.51 (1.39–4.18)**
2 99 4.20 55 5.47 1.30 (1.19–1.37)* 2.87 (1.85–4.14)*
3 12 1.51 4 2.26 1.50 (0.87–1.89) 7.94 (0.32–16.43)
4 (lowest) 25 4.58 35 6.63 1.45 (1.39–1.59)*** 2.80 (1.21–6.44)**
Insured premium (NT$)
< 18 000 190 3.75 121 5.25 1.40 (1.34–1.45)*** 3.86 (2.11–4.86)***
18 000–34 999 2 3.08 5 13.14 4.27 (1.44–5.35)** 9.50 (1.89–18.64)*
≥35 000 0 0 0 0 — —

*P < 0.05; **P < 0.01; ***P < 0.001.

†
aHR: multiple analyses, including age, sex, co-morbidities and socio-economic status.
1 US$, 31.5 NT$ in 2016; aHR, adjusted hazard ratio; CKD, chronic kidney disease; CVD, cardiovascular disease; DM, diabetes melli-
tus; HCV, hepatitis C virus; IRR, incidence rate ratio; PAD, peripheral arterial disease; rate, incidence rate (per 1000 person-years); TB,
tuberculosis.

© 2017 Asian Pacific Society of Respirology Respirology (2017) 22, 1670–1676

TB increases the risk of PAD 1675

Table 3 Incidence and aHR for PAD in the TB and non-TB cohorts stratified by TB type

Subgroup Event Rate IRR (95% CI) aHR† (95% CI)

Non-TB 192 3.73 1.00 (reference) 1.00 (reference)

Pulmonary TB 110 5.58 1.49 (1.42–1.56)*** 3.90 (2.94–4.86)***
Miliary TB 2 5.74 1.54 (0.68–3.21) 2.56 (0.68–10.49)
Extrapulmonary TB 17 4.35 1.16 (1.01–1.50)*** 2.11 (1.31–3.56)**

**P < 0.01; ***P < 0.001.

†
aHR: multiple analyses, including age, sex, co-morbidities and socio-economic status.
aHR, adjusted hazard ratio; IRR, incidence rate ratio; PAD, peripheral arterial disease; rate, incidence rate (per 1000 person-years);
TB, tuberculosis.

Table 4 Incidence and aHR for PAD in the TB and non-TB cohorts stratified by follow-up year

TB

No Yes Compared with the non-TB cohort

Variables Event Rate Event Rate IRR (95% CI) aHR† (95% CI)

Follow-up time
<6 months 28 2.25 37 5.33 2.37 (0.96–4.77) 1.02 (0.49–2.28)
6 months–1 year 15 1.12 16 2.30 2.06 (0.98–3.45) 1.05 (0.31–4.07)
1–5 years 44 4.40 55 7.77 1.77 (1.72–1.85)*** 3.45 (2.00–4.62)***
>5 years 105 6.75 18 7.20 1.07 (1.01–1.31)* 2.86 (1.04–3.98)*

*P < 0.05; ***P < 0.001.

†
aHR: multiple analyses, including age, sex, co-morbidities and socio-economic status.
aHR, adjusted hazard ratio; IRR, incidence rate ratio; PAD, peripheral arterial disease; rate, incidence rate (per 1000 person-years);
TB, tuberculosis.

system restrains tubercle bacilli during LTBI.46 The sim- monitor the development of PAD when caring for
ilarity of inflammatory cytokines and immune cells in patients with TB.
the pathogenesis in LTBI and atherosclerosis may
imply that the development of PAD in patients with
active TB begins during LTBI. Approximately 2–3 bil- Acknowledgements
lion people worldwide have LTBI, and 5–15% of them The present study was based in part on data from the NHIRD,
have developed active TB.47 Studies to investigate the provided and managed by the National Health Research Insti-
tutes, Taiwan. The interpretations and conclusions in this article
role of LTBI in PAD development may be warranted.
do not represent the views of the Bureau of NHI, the Department
This study had some limitations. First, the insurance of Health or the National Health Research Institutes.
data lacked information about the patients’ behaviour,
such as alcohol abuse and substance use, and other
risk factors for PAD, such as smoking, obesity, dietary REFERENCES
habits and the current use of medications, including
anticoagulant agents and hormone replacement ther- 1 World Health Organization. Global Tuberculosis Report. Geneva:
apy; these variables might have influenced our primary World Health Organization, 2016.
outcomes. Therefore, we adjusted for several smoking- 2 Centers for Disease Control. Taiwan Tuberculosis Control Report
related co-morbidities, such as COPD, asthma, IHD 2013. Taipei: Ministry of Health and Welfare, 2014.
3 Philips JA, Ernst JD. Tuberculosis pathogenesis and immunity.
and stroke, as well as obesity-related diseases, such as
Annu. Rev. Pathol. 2012; 7: 353–84.
DM, hypertension and hyperlipidaemia, to minimize 4 Djoba Siawaya JF, Beyers N, van Helden P, Walzl G. Differential
the potential confounding effects. Second, pulmonary cytokine secretion and early treatment response in patients with
images, acid fast staining and culture results were not pulmonary tuberculosis. Clin. Exp. Immunol. 2009; 156: 69–77.
available; thus, we could not assess the severity of TB 5 Smith I. Mycobacterium tuberculosis pathogenesis and molecular
or confirm the timing of smear- or culture-negative determinants of virulence. Clin. Microbiol. Rev. 2003; 16: 463–96.
results in patients with TB. Third, the effects of and 6 Kaufmann SH, Dorhoi A. Inflammation in tuberculosis: interac-
treatment periods with different anti-TB medications tions, imbalances and interventions. Curr. Opin. Immunol. 2013;
were not evaluated. Fourth, bias might remain in this 25: 441–9.
retrospective cohort study despite the meticulous 7 Narasimhan P, Wood J, MacIntyre CR, Mathai D. Risk factors for
tuberculosis. Pulm. Med. 2013; 2013: 828939.
adjustment of the model for potential confounders.
8 Abdelrahman M, Sinha AK, Karkar A. Tuberculosis in end-stage
In conclusion, patients with TB had a 3.93-fold renal disease patients on hemodialysis. Hemodial. Int. 2006; 10:
higher risk of developing PAD than the general popula- 360–4.
tion using data from a nationwide population-based 9 Hnizdo E, Murray J. Risk of pulmonary tuberculosis relative to sili-
database, and this risk became significant after a 1-year cosis and exposure to silica dust in South African gold miners.
follow-up period. We recommend that physicians Occup. Environ. Med. 1998; 55: 496–502.

Respirology (2017) 22, 1670–1676 © 2017 Asian Pacific Society of Respirology

1676 S-H Wang et al.

10 Peto HM, Pratt RH, Harrington TA, LoBue PA, Armstrong LR. Epi- review of the literature. Clin. Physiol. Funct. Imaging 2012; 32:
demiology of extrapulmonary tuberculosis in the United States, 421–30.
1993-2006. Clin. Infect. Dis. 2009; 49: 1350–7. 33 Iriz E, Cirak MY, Engin ED, Zor MH, Erer D, Imren Y, Turet S,
11 Lin YC, Liang SJ, Liu YH, Hsu WH, Shih CM, Sung FC, Chen W. Halit V. Effects of atypical pneumonia agents on progression of
Tuberculosis as a risk factor for systemic lupus erythematosus: atherosclerosis and acute coronary syndrome. Acta Cardiol. 2007;
results of a nationwide study in Taiwan. Rheumatol. Int. 2012; 32: 62: 593–8.
1669–73. 34 Cerrato E, D’Ascenzo F, Biondi-Zoccai G, Omedè P, Moretti C,
12 Chung WS, Lin CL, Hung CT, Chu YH, Sung FC, Kao CH, Yeh JJ. Cicalini S, Parthasarathi G, Sheiban I, Gaita F. Acute coronary syn-
Tuberculosis increases the subsequent risk of acute coronary syn- drome in HIV patients: from pathophysiology to clinical practice.
drome: a nationwide population-based cohort study. Int. J. Tuberc. Cardiovasc. Diagn. Ther. 2012; 2: 50–5.
Lung Dis. 2014; 18: 79–83. 35 Van Eeden S, Leipsic J, Paul Man SF, Sin DD. The relationship
13 Peng YC, Lin CL, Hsu WY, Chang CS, Yeh HZ, Kao CH. Risk of between lung inflammation and cardiovascular disease. Am.
liver cirrhosis in patients with tuberculosis: a nationwide cohort J. Respir. Crit. Care Med. 2012; 186: 11–6.
study. Eur. J. Clin. Invest. 2015; 45: 663–9. 36 Zhu J, Katz RJ, Quyyumi AA, Canos DA, Rott D, Csako G, Zalles-
14 Shen CH, Chou CH, Liu FC, Lin TY, Huang WY, Wang YC, Ganley A, Ogunmakinwa J, Wasserman AG, Epstein SE. Associa-
Kao CH. Association between tuberculosis and Parkinson disease: tion of serum antibodies to heat-shock protein 65 with coronary
a nationwide, population-based cohort study. Medicine (Baltimore) calcification levels: suggestion of pathogen-triggered autoimmunity
2016; 95: e2883. in early atherosclerosis. Circulation 2004; 109: 36–41.
15 Shammas NW. Epidemiology, classification, and modifiable risk 37 Wick G, Jakic B, Buszko M, Wick MC, Grundtman C. The role of
factors of peripheral arterial disease. Vasc. Health Risk Manag. heat shock proteins in atherosclerosis. Nat. Rev. Cardiol. 2014; 11:
2007; 3: 229–34. 516–29.
16 Fowkes FGR, Rudan D, Rudan I, Aboyans V, Denenberg JO, 38 Turken O, Kunter E, Sezer M, Solmazgul E, Cerrahoglu K,
McDermott MM, Norman PE, Sampson UKA, Williams LJ, Bozkanat E, Ozturk A, Ilvan A. Hemostatic changes in active pul-
Mensah GA et al. Comparison of global estimates of prevalence monary tuberculosis. Int. J. Tuberc. Lung Dis. 2002; 6: 927–32.
and risk factors for peripheral artery disease in 2000 and 2010: a 39 Mert A, Bilir M, Tabak F, Ozaras R, Ozturk R, Senturk H, Aki H,
systematic review and analysis. Lancet 2013; 382: 1329–40. Seyhan N, Karayel T, Aktuglu Y. Miliary tuberculosis: clinical mani-
17 Joosten MM, Pai JK, Bertoia ML, Rimm EB, Spiegelman D, festations, diagnosis and outcome in 38 adults. Respirology 2001; 6:
Mittleman MA, Mukamal KJ. Associations between conventional 217–24.
cardiovascular risk factors and risk of peripheral artery disease in 40 Guilford T, Morris D, Gray D, Venketaraman V. Atherosclerosis:
men. JAMA 2012; 308: 1660–7. pathogenesis and increased occurrence in individuals with HIV
18 Hiatt WR, Hoag S, Hamman RF. Effect of diagnostic criteria on the and Mycobacterium tuberculosis infection. HIV AIDS (Auckl.) 2010;
prevalence of peripheral arterial disease. The San Luis Valley Dia- 2: 211–8.
betes Study. Circulation 1995; 91: 1472–9. 41 Sharma SK, Mohan A, Sharma A, Mitra DK. Miliary tuberculosis:
19 Criqui MH, Aboyans V. Epidemiology of peripheral artery disease. new insights into an old disease. Lancet Infect. Dis. 2005; 5:
Circ. Res. 2015; 116: 1509–26. 415–30.
20 Bloemenkamp DG, van den Bosch MA, Mali WP, Tanis BC, 42 Camargo CA, Stampfer MJ, Glynn RJ, Gaziano JM, Manson JE,
Rosendaal FR, Kemmeren JM, Algra A, Visseren FL, van der Goldhaber SZ, Hennekens CH. Prospective study of moderate
Graaf Y. Novel risk factors for peripheral arterial disease in young alcohol consumption and risk of peripheral arterial disease in US
women. Am. J. Med. 2002; 113: 462–7. male physicians. Circulation 1997; 95: 577–80.
21 Packard RRS, Libby P. Inflammation in atherosclerosis: from vas- 43 Rajagopalan S, Dellegrottaglie S, Furniss AL, Gillespie BW,
cular biology to biomarker discovery and risk prediction. Clin. Satayathum S, Lameire N, Saito A, Akiba T, Jadoul M, Ginsberg N
Chem. 2008; 54: 24–38. et al. Peripheral arterial disease in patients with end-stage renal
22 Libby P. Inflammation in atherosclerosis. Nature 2002; 420: disease. Circulation 2006; 114: 1914–22.
868–74. 44 Periard D, Cavassini M, Taffé P, Chevalley M, Senn L, Chapuis-
23 Libby P, Ridker PM, Maseri A. Inflammation and atherosclerosis. Taillard C, de Vallière S, Hayoz D, Tarr PE. High prevalence of
Circulation 2002; 105: 1135–43. peripheral arterial disease in HIV-infected persons. Clin. Infect.
24 Brevetti G, Giugliano G, Brevetti L, Hiatt WR. Inflammation in Dis. 2008; 46: 761–7.
peripheral artery disease. Circulation 2010; 122: 1862–75. 45 Bentzon JF, Otsuka F, Virmani R, Falk E. Mechanisms of plaque
25 Flynn JL, Chan J, Lin PL. Macrophages and control of granulomatous formation and rupture. Circ. Res. 2014; 114: 1852–66.
inflammation in tuberculosis. Mucosal Immunol. 2011; 4: 271–8. 46 Dutta NK, Karakousis PC. Latent tuberculosis infection: myths,
26 Sasindran SJ, Torrelles JB. Mycobacterium tuberculosis infection models, and molecular mechanisms. Microbiol. Mol. Biol. Rev.
and inflammation: what is beneficial for the host and for the bacte- 2014; 78: 343–71.
rium? Front. Microbiol. 2011; 2: 2. 47 Ai JW, Ruan QL, Liu QH, Zhang WH. Updates on the risk factors
27 Huaman MA, Henson D, Ticona E, Sterling TR, Garvy BA. Tuber- for latent tuberculosis reactivation and their managements. Emerg.
culosis and cardiovascular disease: linking the epidemics. Trop. Microbes Infect. 2016; 5: e10.
Dis. Travel Med. Vaccines 2015; 1: 10.
28 Cheng CL, Kao YH, Lin SJ, Lee CH, Lai ML. Validation of the
National Health Insurance Research Database with ischemic stroke Supplementary Information
cases in Taiwan. Pharmacoepidemiol. Drug Saf. 2011; 20: 236–42. Additional supplementary information can be accessed via the
29 Cheng CL, Lee CH, Chen PS, Li YH, Lin SJ, Yang YH. Validation of html version of this article at the publisher’s website.
acute myocardial infarction cases in the National Health Insurance
Research Database in Taiwan. J. Epidemiol. 2014; 24: 500–7. Figure S1 Flow chart of the process used to select the
30 Yeh CC, Wang HH, Chou YC, Hu CJ, Chou WH, Chen TL, Liao CC. study subjects from the National Health Insurance
High risk of gastrointestinal hemorrhage in patients with epilepsy: Research Database in Taiwan.
a nationwide cohort study. Mayo Clin. Proc. 2013; 88: 1091–8.
Table S1 Classification of tuberculosis in ICD-9-CM.
31 Shen CH, Lin TY, Huang WY, Chen HJ, Kao CH. Pneumoconiosis
increases the risk of peripheral arterial disease: a nationwide
Table S2 Definitions of peripheral arterial disease with
population-based study. Medicine (Baltimore) 2015; 94: e911. ICD-9-CM codes used in this study.
32 Roed T, Lebech AM, Kjaer A, Weis N. Hepatitis C virus Table S3 Adjusted hazard ratio for peripheral arterial
infection and risk of coronary artery disease: a systematic disease based on different risk factors.

© 2017 Asian Pacific Society of Respirology Respirology (2017) 22, 1670–1676