RESEARCH ARTICLE

Association of iron deficiency anemia with

tuberculosis in Taiwan: A nationwide
population-based study
Kuo-An Chu1,2,3☯, Chun-Hsiang Hsu1☯, Mei-Chen Lin4, Yi-Hsin Chu5, Yao-
Min Hung ID3,6,7,8*, James Cheng-Chung Wei9,10,11,12*
1 Division of Chest Medicine, Department of Internal Medicine, Kaohsiung Veterans General Hospital,
Kaohsiung, Taiwan, 2 Department of Nursing, Shu-Zen Junior College of Medicine and Management,
Kaohsiung, Taiwan, 3 School of Medicine, National Yang Ming University, Taipei, Taiwan, 4 Management
a1111111111 Office for Health Data, China Medical University Hospital, Taichung, Taiwan, 5 Department of Family
a1111111111 Medicine, National Cheng Kung University Hospital, Tainan, Taiwan, 6 Kaohsiung Municipal United Hospital,
a1111111111 Kaohsiung, Taiwan, 7 Department of Emergency Medicine, Kaohsiung Veterans General Hospital,
a1111111111 Kaohsiung, Taiwan, 8 Yuhing Junior College of Health Care and Management, Kaohsiung, Taiwan,
a1111111111 9 Department of Rheumatology, BenQ Medical Center, The Affiliated BenQ Hospital of Nanjing Medical
University, China, 10 Institute of Medicine, Chung Shan Medical University, Taichung, Taiwan,
11 Department of Medicine, Chung Shan Medical University Hospital, Taichung, Taiwan, 12 Graduate
Institute of Integrated Medicine, China Medical University, Taichung, Taiwan

☯ These authors contributed equally to this work.

OPEN ACCESS * ymhung1@gmail.com (YMH); jccwei@gmail.com (JCCW)

Citation: Chu K-A, Hsu C-H, Lin M-C, Chu Y-H,

Hung Y-M, Wei JC-C (2019) Association of iron
deficiency anemia with tuberculosis in Taiwan: A Abstract
nationwide population-based study. PLoS ONE 14
(8): e0221908. https://doi.org/10.1371/journal.
Background
pone.0221908
Iron deficiency is associated with decreased cellular immunity, which may predispose
Editor: Wisit Cheungpasitporn, University of
Mississippi Medical Center, UNITED STATES patients with iron deficiency anemia (IDA) to increased risk of developing tuberculosis (TB).
This study investigated the relationship between newly diagnosed IDA and TB infection in
Received: May 28, 2019
Taiwan.
Accepted: August 16, 2019

Published: August 30, 2019 Methods

Copyright: © 2019 Chu et al. This is an open The study included data on 21,946 patients with incident IDA and 87,555 non-IDA controls
access article distributed under the terms of the
from a national database covering the period 2000–2012. IDA and non-IDA subjects were
Creative Commons Attribution License, which
permits unrestricted use, distribution, and matched 1:4 on age, gender, and index year. The follow-up period was defined as the time
reproduction in any medium, provided the original from the initial IDA diagnosis to the date of developing TB or 31 December 2013. Cox pro-
author and source are credited. portional hazards models were used to estimate hazard ratios and 95% confidence inter-
Data Availability Statement: All relevant data are vals, with the control group as the reference.
within the manuscript.

Funding: This work was supported by grants from Results

the Ministry of Health and Welfare, Taiwan
(MOHW108-TDU-B-212-133004), China Medical
The adjusted hazard ratio of TB for the IDA group was 1.99 (95% confidence interval, 1.77–
University Hospital, Academia Sinica Stroke 2.25) compared with the control group. The subgroup analysis showed that for both gen-
Biosignature Project (BM10701010021), MOST ders, all age groups, and patients with diabetes mellitus, hyperlipidemia, hypertension, can-
Clinical Trial Consortium for Stroke (MOST 107-
cer, chronic obstructive pulmonary disease, and hepatitis B virus infection, the IDA group
2321-B-039 -004-), Tseng-Lien Lin Foundation,
Taichung, Taiwan, and Katsuzo and Kiyo Aoshima had significantly higher TB incidence. The association was significantly stronger within the 5
Memorial Funds, Japan. years after new IDA diagnosis for both genders and all age groups.

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 1 / 11

 Association of iron deficiency anemia with TB

Competing interests: The authors have declared Conclusions

that no competing interests exist.
Higher TB incidence was discovered in the IDA group, especially for patients with
comorbidities.

Introduction
Tuberculosis (TB), an infectious disease caused by Mycobacteria tuberculosis, is characterized by
tubercle bacilli displaying intracellular survival strategies and chronic pulmonary inflammation
[1]. The global burden of TB is still increasing, and the morbidity and mortality of TB remain
substantial [2]. In 2012, approximately 8.6 million new cases of TB and an estimated 1.3 million
related mortalities were reported globally [3]. TB affected approximately 10.4 million people
and caused 1.7 million deaths worldwide in 2016 [4]. Preventing TB infection is thus a crucial
global health issue. TB is endemic and highly prevalent in Taiwan. In 2012, the incidences of TB
and TB-related death were 53 and 2.7 cases per 100,000 individuals, respectively [5]. Researchers
have been searching for more means of preventing new cases of TB [2]. An assessment of poten-
tially modifiable risk factors is a promising consideration for the formulation of TB control poli-
cies [6]. Some TB risk factors have been known for decades, including systemic diseases such as
diabetes mellitus (DM) [7] and chronic kidney disease [8] as well as tobacco smoking [9], alco-
hol use [10], body mass index [11], silicosis [12], human immunodeficiency virus (HIV) infec-
tion [13], splenectomy [14], and gastrectomy [15,16]. Most of these risk factors can impair the
human immune system, thereby increasing the risk of TB. However, few studies have investi-
gated the link between nutritional iron deficiency and high TB prevalence.
Micronutrient deficiencies are a well-known global health threat, and poor nutritional sta-
tus may predispose an individual to some infectious diseases [17]. Anemia, a critical global
health problem, is the most common micronutrient deficiency, occurring in approximately
one quarter of the world’s population [17,18]. Iron deficiency anemia (IDA) is the most signifi-
cant contributor, accounting for 50% of all cases of anemia. IDA has a prevalence of 2%–5%
among adult men and postmenopausal women in the developed world [19]. Animal and
human studies have demonstrated that nutritional iron deficiency is associated with impaired
phytohemagglutinin-induced lymphocyte proliferation and delayed-type hypersensitivity
responses with relative preservation of humoral immunity [20–25].
TB and IDA are major public health concerns worldwide. Nonetheless, the relationship
between iron deficiency, especially IDA, and the risk of contracting TB remains unclear. By
using data in the Taiwan National Health Insurance Research Database (NHIRD) for the
period January 1, 2000, to December 31, 2012 in the current population-based cohort study,
we examined the association between newly diagnosed IDA and subsequent TB development.

Material and methods

Study design
This study is a retrospective matched-cohort study analyzing data over a period of 12 years in
a nationwide, population-based database in Taiwan.

Database
The IDA and control cohorts were created using the Taiwan NHIRD, which is maintained by
a single national health insurance (NHI) program that covered 99.6% of Taiwan’s population

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 2 / 11

 Association of iron deficiency anemia with TB

until 2011. The NHIRD contains patient demographic information, encrypted identification
number, gender, birth date, diagnostic data and procedures, all types of medical visits (outpa-
tient department, emergency care, and hospitalizations), traditional Chinese medical services,
and prescription drug use. The diagnostic and procedure codes are based on the International
Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). The Longitudi-
nal Health Insurance Database 2000 (LHID 2000), a subset of the NHIRD, was employed in
this study. The LHID 2010 is composed of all original claims data of 1,000,000 randomly sam-
pled beneficiaries of the NHI program. No significant differences in age distribution, gender,
or health care cost were noted between the 1,000,000 people in the LHID 2010 and the individ-
uals in the NHIRD.

Study population
Cases of IDA and TB were identified from the NHIRD by using the corresponding ICD-9
codes 280.X and 011–018, respectively, for the period from January 2000 to December 2012.
We first identified patients with newly diagnosed IDA from data concerning both outpatient
and inpatient visits. IDA was defined through diagnostic ICD codes and procedure codes,
including complete blood count tests and serum ferritin tests. The index date was defined as
the first date of IDA diagnosis. For further ascertainment, only patients with at least one inpa-
tient admission or three outpatient visits during the 1 year after IDA was first diagnosed were
selected. We then excluded patients who had previous history of TB (ICD-9-CM 010.x to 018.
x and anti-TB drugs for 2 months), those <20 years of age, and those who withdrew from the
insurance program before the index date. The IDA group finally comprised 21,946 patients.
The non-IDA control group was randomly selected from the patients who had never been
diagnosed with IDA or TB (selected at a 1:4 ratio matched by age, gender, and index year).
The non-IDA control group comprised 87,555 people. Individuals in both groups were tracked
until a TB event, withdrawal from the NHI program, or the end of 2013, whichever occurred
first.

Covariables
Factors that might influence the incidence of TB—such as age, gender, income level, and
comorbidities—were used as independent variables. We classified age into three groups: 20–
39, 40–64, and �65 years. The comorbidities analyzed in this study were hypertension (ICD-
9-CM codes 401–405), DM (ICD-9-CM code 250), hyperlipidemia (ICD-9-CM code 272),
chronic obstructive pulmonary disease (COPD; ICD-9-CM codes 491, 492, 496), cancer (ICD-
9-CM codes 140–208), chronic kidney disease (CKD) (ICD-9-CM code 585),alcoholic liver
disease (ICD-9-CM 571.0, 571.1, 571.3), liver cirrhosis (ICD-9-CM code 571.4), hepatitis B
(ICD-9-CM codes 070.2, 070.3, V02.61), hepatitis C (ICD-9-CM codes 070.41, 070.44, 070.51,
070.54, V02.62), HIV infection (ICD-9-CM code 042–044 795.8 V08), pneumoconiosis (ICD-
9-CM code 042–044 795.8 V08), splenectomy (ICD-9-CM procedure code 41.5), partial gas-
trectomy (ICD-9-OP 43.5, 43.6, 43.7, 43.8, 43.81, 43.82, 43.89), and total gastrectomy (ICD-
9-OP 43.91 and 43.99). Information on comorbidities was obtained by tracing all ambulatory
medical care and inpatient records in the NHI database for the 2 years before the index date.

Outcome measurement
Regarding outcomes, we focused on the development of Mycobacterium tuberculosis infection.
Diagnosis of TB was identified using ICD-9-CM codes 010–018 in combination with the pre-
scription of at least two anti-TB drugs within 6 months of TB diagnosis.

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 3 / 11

 Association of iron deficiency anemia with TB

Statistical analysis
Proportional differences in independent variables between the IDA and control cohorts were
analyzed using the Pearson χ2 test. The incidence of TB was expressed as the number of newly
diagnosed TB cases per 10,000 person-years. To assess the risk of subsequently developing TB,
we performed Cox regression analysis to obtain the crude and adjusted hazards ratios (HRs)
and 95% confidence intervals (CIs) for the case group compared with the control group. Cox
regression models were adjusted for age, gender, and all comorbidities. Furthermore, we per-
formed stratified analysis by calculating the HRs for the patients with IDA according to differ-
ent subgroups. The significance level was set to a two-tailed p value of 0.05. All data analyses
were performed using SAS1 (version 9.4; SAS Institute, Inc., Cary, NC, USA).
This study was approved by the Institutional Review Board of China Medical University
(permit number: CMUH-104-REC2-115-R3). Because all data were used anonymously and
fully deidentified before analysis, the need for informed consent was waived by the board.

Results
Overall, 413 individuals in the IDA group (which comprised 21,946 patients) developed newly
onset TB during the follow-up period. The longest follow-up period was 8 years. Overall, the
crude TB infection rate, which is the crude HR (95% CI), was 1.94 (1.73–2.17) for the IDA
group.

Demographic characteristics and comorbidities of patients with newly

diagnosed IDA and the comparison cohort in Taiwan during 2000–2012
Table 1 displays the clinical characteristics of the patients with and without IDA. IDA was
more prevalent in women than men (72.3% vs. 27.7%). The mean age at time of IDA diagnosis
was 53.6 years. Comorbidities were more common in the IDA group.

Risk factors of new-onset TB among the patients with IDA and comparison
cohort
Table 2 presents the results of univariate and multivariate Cox regression analyses. Among all
of the relevant variables, being in the IDA cohort, age > 40 years, male gender, gastrectomy,
and comorbidities of DM, chronic kidney disease, COPD, pneumoconiosis, liver cirrhosis
were the risk factors associated with a higher risk of developing TB. The crude HR (95% CI)
for the IDA group was 1.94 (1.73–2.17). After adjustment for age, gender, and comorbidities,
the risk of developing TB was similar (aHR, 1.99; 95% CI, 1.77–2.25). The survival curve (Fig
1) shows that the cumulative incidence of TB was higher in the IDA cohort than in the com-
parison group (p < 0.0001).

Incidence rate, HR, and CI of TB in different subgroups

Table 3 reveals the association of risk of TB between gender, age, and comorbidity subgroups.
In the gender-subgroup analysis, both men and women with IDA had higher TB risk than
those without IDA. In all age groups, the IDA group had a higher risk of TB than the compari-
son group. In different stratifications of baseline comorbidities, the patients with DM, hyper-
lipidemia, hypertension, cancer, COPD, or hepatitis B virus infection in the IDA group had a
significantly higher risk of TB. For the patients with hepatitis B virus infection, the IDA group
had a 5.98-fold higher risk of TB (95% CI, 2.5–14.32).

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 4 / 11

 Association of iron deficiency anemia with TB

Table 1. Demographic characteristics and comorbidities of patients with newly diagnosed iron deficiency anemia in Taiwan during 2000–2012.
Characteristics Total Iron deficiency anemia p value
No Yes
n = 87,555 n = 21,946
n n (%) / mean ± SD n (%) / mean ± SD
Gender
Female 79182 63313 (72.3) 15869 (72.3) 0.99
Male 30319 24242 (27.7) 6077 (27.7)
Age
20–39 27025 21620 (24.7) 5405 (24.63) 0.88
40–64 48615 38892 (44.42) 9723 (44.30)
�65 33861 27043 (30.89) 6818 (31.07)
Mean (SD) a 53.55±18.29 53.72±18.40 0.22
Baseline comorbidity
Hypertension 37518 28550 (32.6) 8968 (40.9) <0.001
Diabetes mellitus 18896 13776 (15.7) 5120 (23.3) <0.001
Hyperlipidemia 25005 19250 (22.0) 5755 (26.2) <0.001
Chronic kidney disease 3143 1616 (1.8) 1527 (7.0) <0.001
Cancer 39862 29043 (33.2) 10819 (49.3) <0.001
COPD 16293 12280 (14.0) 4013 (18.3) <0.001
Alcoholic liver disease 900 479 (0.5) 421 (1.9) <0.001
Liver cirrhosis 2071 1007 (1.2) 1064 (4.8) <0.001
Hepatitis B 3294 2398 (2.7) 896 (4.1) <0.001
Hepatitis C 1668 1075 (1.2) 593 (2.7) <0.001
HIV infection 32 18 (0.0) 14 (0.1) <0.001
Pneumoconiosis 1149 780 (0.9) 369 (1.7) <0.001
Splenectomy 90 52 (0.1) 38 (0.2) <0.001
Gastrectomy 360 187 (0.2) 173 (0.8) <0.001
a
Chi-square test, t-test

https://doi.org/10.1371/journal.pone.0221908.t001

Risk of tuberculosis according to follow-up years by age group

Table 4 details the incidence and HR of TB stratified by follow-up years. In the analysis of the
three age subgroups, the aHR (95% CI) of TB for the IDA group was the highest and most sig-
nificant in the first 2 years, at 3.33 (1.61–6.88), 2.59 (1.75–3.84), and 2.15 (1.72–2.69) in the
20–39, 40–64, and �65 years age groups, respectively (p < 0.001). In all age subgroups, IDA
was associated with higher incidence of subsequent TB, especially within the first 2 years after
IDA diagnosis; even 5 years after IDA diagnosis, these patients were more likely to develop
TB. In the 40–64 years age subgroup, the patients with IDA exhibited a significantly increased
association with TB compared with the comparison group even after �5 follow-up years.

Discussion
In this nationwide population-based study of data covering 12 years, we discovered that people
with new diagnoses of IDA were nearly twice as likely to subsequently develop TB than were
those without IDA. Overall, IDA was associated with a 99% increased incidence of TB com-
pared with the matched group. This result supports the hypothesis that individuals with iron
deficiency are more susceptible to infections, perhaps because of impaired cell-mediated
immunity. Furthermore, the effects of IDA were found to be more significant in some high-

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 5 / 11

 Association of iron deficiency anemia with TB

Table 2. Cox-model–measured hazard ratio and 95% confidence interval of tuberculosis associated with iron deficiency anemia in patients.
Characteristics Event Person IR Crude Adjusted a
(n = 1361) year HR (95% CI) p value HR (95% CI) p value
Iron deficiency anemia
No 948 589070 16.09 Ref. Ref.
Yes 413 131327 31.45 1.94(1.73–2.17) <0.001 1.99(1.77–2.25) <0.001
Gender
Female 624 547074 11.41 Ref. Ref.
Male 737 173322 42.52 3.64(3.28–4.06) <0.001 2.04(1.83–2.29) <0.001
Age at baseline
20–39 85 201705 4.21 Ref. Ref.
40–64 359 342640 10.48 2.48(1.95–3.14) <0.001 2.21(1.74–2.81) <0.001
�65 917 176051 52.09 12.03(9.62–15.03) <0.001 7.55(5.9–9.65) <0.001
Baseline comorbidity
Hypertension 773 207254 37.30 3.16(2.84–3.52) <0.001 1.02(0.9–1.17) 0.75
Diabetes mellitus 398 100892 39.45 2.45(2.18–2.76) <0.001 1.23(1.08–1.41) 0.002
Hyperlipidemia 398 141213 28.18 1.64(1.46–1.85) <0.001 0.8(0.7–0.91) <0.001
Chronic kidney disease 89 13095 67.96 3.54(2.85–4.39) <0.001 1.38(1.1–1.72) 0.01
Cancer 441 240071 18.37 0.94(0.84–1.05) 0.27 0.89(0.79–1) 0.05
COPD 479 82567 58.01 4.05(3.62–4.53) <0.001 1.75(1.55–1.98) <0.001
Alcoholic liver disease 20 4254 47.02 2.38(1.53–3.70) <0.001 1.38(0.87–2.17) 0.17
Liver cirrhosis 56 8448 66.29 3.38(2.58–4.42) <0.001 1.53(1.15–2.05) 0.004
Hepatitis B 24 16922 14.18 0.71(0.48–1.07) 0.10 0.62(0.41–0.94) 0.02
Hepatitis C 34 7658 44.40 2.25(1.60–3.16) <0.001 1.15(0.8–1.64) 0.46
HIV infection 1 139 71.79 3.54(0.50–25.00) 0.20 2.77(0.39–19.77) 0.31
Pneumoconiosis 43 4666 92.16 4.72(3.48–6.40) <0.001 1.66(1.22–2.25) 0.001
Splenectomy 2 364 54.87 2.71(0.68–10.83) 0.16 1.34(0.33–5.44) 0.68
Gastrectomy 16 1699 94.15 4.81(2.94–7.88) <0.001 1.74(1.05–2.87) 0.03

HR, hazard ratio; CI, confidence interval.

a
Adjusted HR: adjusted for gender, age, and all comorbidities in the Cox proportional hazards regression.

https://doi.org/10.1371/journal.pone.0221908.t002

risk groups, such as patients with DM, hyperlipidemia, hypertension, cancer, COPD, or hepa-
titis B virus infection. Age is also another consideration for the prevention of TB among those
with newly diagnosed IDA. Age had effects on both the strength and duration of the TB
association.
To the best of our knowledge, this is the first large-scale study to assess the association of
newly diagnosed IDA with subsequent TB. Our study is unique for several reasons. First, the
sample size of the current study was large (enrolling 109,501 patients overall) and the follow-
up period (131,326.7 person-years for the IDA cohort) was longer than in all other studies
investigating the incidence of TB for various subgroups of patients with anemia. This strength
enabled us to analyze the association of IDA with TB in groups stratified by age, gender,
comorbidities, and follow-up period. Second, our data were obtained from a national insur-
ance database covering a whole country with a single ethnic population; thus, the results are
superior to those in smaller, single-hospital, and specific age or gender studies and those using
purposive sampling [26,27]. This strength minimized potential bias from the sampling process
[28]. Third, we defined IDA and TB by using accurate diagnosis criteria. IDA cases were
defined using both ICD codes and procedure codes, whereas TB diagnosis was defined using
both ICD codes and prescription codes for at least two anti-TB drugs.

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 6 / 11

 Association of iron deficiency anemia with TB

Fig 1. Cumulative incidence of TB in the IDA cohort and comparison group.

https://doi.org/10.1371/journal.pone.0221908.g001

The study data revealed that compared with the general population, patients with IDA had
a greater incidence of subsequent TB development. The overall TB risk of the patients with
IDA was higher than that of general population (aHR, 1.99 [95% CI, 1.77–2.25]), with an aHR
(95% CI) of 2.02 (1.72–2.39) and 1.96 (1.64–2.33) for men and women, respectively. In addi-
tion, we identified some high-TB-risk groups of patients with IDA. Patients with DM, hyper-
lipidemia, hypertension, cancer, COPD, or hepatitis B virus infection in the IDA group were
significantly more likely to develop TB. For the patients with hepatitis B virus infection, the
IDA group had a 5.98-fold higher association with TB (95% CI, 2.5–14.32).
Another finding of this study that deserves attention is the effect of age on the TB associa-
tion and the duration of the association. For all age groups, the IDA group had a stronger asso-
ciation with TB than the comparison group. However, as age increased, the association with
TB infection became weaker. The aHR (95% CI) of TB association decreased from 2.75 (1.75–

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 7 / 11

 Association of iron deficiency anemia with TB

Table 3. Incidence rates, hazard ratio, and confidence interval of tuberculosis in different stratifications.
Variables Control Case Case VS. Control
n = 87,555 n = 21,946 Crude HR Adjusted HR a
Event Person years IR Event Person years IR (95% CI) (95% CI)
Overall 948 589070 16.09 413 131327 31.45 1.94(1.73–2.17) c 1.99(1.77–2.25) c
Gender
Female 426 443558 9.60 198 103516 19.13 1.98(1.68–2.35) c 1.96(1.64–2.33) c
c
Male 522 145512 35.87 215 27811 77.31 2.11(1.80–2.47) 2.02(1.72–2.39) c
Age at baseline
20–39 48 161804 2.97 37 39901 9.27 3.12(2.03–4.80) c 2.75(1.75–4.33) c
c
40–64 228 278791 8.18 131 63850 20.52 2.50(2.02–3.11) 2.11(1.68–2.65) c
c
�65 672 148475 45.26 245 27576 88.84 1.93(1.66–2.23) 1.83(1.57–2.13) c
Baseline comorbidity
Hypertension 532 164495 32.34 241 42759 56.36 1.71(1.47–2.00) c 1.76(1.51–2.06) c
c
Diabetes mellitus 258 77158 33.44 140 23734 58.99 1.73(1.41–2.12) 1.77(1.43–2.19) c
c
Hyperlipidemia 277 111779 24.78 121 29434 41.11 1.64(1.33–2.03) 1.67(1.34–2.08) c
Chronic kidney disease 42 6846 61.35 47 6249 75.21 1.25(0.83–1.90) 1.34(0.87–2.06)
Cancer 285 178190 15.99 156 61881 25.21 1.57(1.30–1.91) c 1.78(1.46–2.18) c
c
COPD 335 64860 51.65 144 17707 81.32 1.56(1.28–1.89) 1.68(1.37–2.05) c
Alcoholic liver disease 8 2507 31.91 12 1746 68.71 2.25(0.92–5.52) 2.09(0.79–5.52)
Liver cirrhosis 23 4650 49.46 33 3798 86.89 1.71(1.01–2.92) b 1.73(0.99–3.04)
c
Hepatitis B 8 12777 6.26 16 4145 38.60 6.17(2.64–14.42) 5.98(2.5–14.32) c
Hepatitis C 21 5272 39.83 13 2386 54.49 1.35(0.68–2.70) 1.42(0.69–2.91)
HIV infection 1 62 161.07 0 77 0.00 - -
Pneumoconiosis 31 3567 86.90 12 1099 109.23 1.21(0.62–2.35) 1.20(0.60–2.40)
Splenectomy 0 247 0.00 2 117 170.89 - -
Gastrectomy 6 870 68.95 10 829 120.59 1.77(0.64–4.88) 1.74(0.57–5.31)

IR, incidence rates, per 10,000 person-years; HR, hazard ratio; CI, confidence interval.
a
Adjusted HR: adjusted for gender, age, and all comorbidities in Cox proportional hazards regression.
b
p < 0.05
c
p < 0.001

https://doi.org/10.1371/journal.pone.0221908.t003

4.33) in the 20–39 years subgroup to 1.83 (1.57–2.13) in the �65 years subgroup. In all age sub-
groups, IDA was associated with higher incidence of subsequent TB, especially in the first 2
years after IDA diagnosis. Even 5 years after IDA diagnosis, these patients were still more likely
to develop TB. However, only in the 40–64 subgroup, patients with IDA exhibited a signifi-
cantly higher association with TB than the comparison group even after �5 follow-up years.
Several reasons may explain why more young patients with IDA developed TB. First, youn-
ger people tend to exhibit more symptoms and thus may be more likely to receive a TB diagno-
sis. Second, it is more likely for older people to be lost to follow-up than younger people with
IDA because several years may have passed since their IDA diagnosis. After several years of fol-
low-up, they may have also been eating iron-rich foods or independently taking drugs rather
than visiting a doctor to obtain such drugs. This may have resulted in older patients with IDA
being less frequently recorded in the NHIRD database.
The present findings have both clinical and public health implications. Clinically, physi-
cians and patients should be aware of the possible association between TB and IDA. When
treating patients with IDA and DM, hyperlipidemia, hypertension, cancer, COPD, or hepatitis
B virus infection, clinicians must be aware of the increased risk of TB incidence. From a public

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 8 / 11

 Association of iron deficiency anemia with TB

Table 4. Incidence and hazard ratio of tuberculosis with stratification by follow-up year.
Variables Control Case Case VS. Control
n = 87,555 n = 21,946 Crude HR Adjusted HR a
Event Person years IR Event Person years IR (95% CI) (95% CI)
Patients less than aged 40 years
<2 17 42128 4.04 16 10454 15.30 3.80(1.92–7.51) d 3.33(1.61–6.88) c
b
2–5 10 51149 1.96 8 12585 6.36 3.25(1.28–8.24) 3.01(1.14–7.97) b
c
�5 21 68527 3.06 13 16862 7.71 2.52(1.26–5.03) 1.98(0.93–4.19)
Patients aged 40–64 years
<2 62 75810 8.18 52 18173 28.61 3.50(2.42–5.06) d 2.59(1.75–3.84) d
d
2–5 74 90030 8.22 46 20757 22.16 2.70(1.87–3.91) 2.34(1.58–3.47) d
b
�5 92 112750 8.16 33 24919 13.24 1.63(1.10–2.43) 1.57(1.04–2.36) b
Patients more than aged 65 years
<2 246 50040 49.16 122 10807 112.89 2.27(1.83–2.82) d 2.15(1.72–2.69) d
d
2–5 221 51345 43.04 77 9370 82.18 1.91(1.47–2.48) 1.81(1.39–2.36) d
b
�5 205 47090 43.53 46 7400 62.16 1.42(1.03–1.96) 1.35(0.97–1.87)

IR, incidence rates, per 10,000 person-years; HR, hazard ratio; CI, confidence interval.
a
Adjusted HR: adjusted for gender, age, and all comorbidities in Cox proportional hazards regression.
b
p < 0.05
c
p < 0.01
d
p < 0.001

https://doi.org/10.1371/journal.pone.0221908.t004

health perspective, policymakers can consider implementing a TB screening test for certain
high-risk patients with IDA.
Some limitations should be noted. Determining a strong association between IDA and TB
by using a diagnosis database is extremely difficult and potentially uncertain. First, the diagno-
ses of IDA and TB in this study were mainly based on diagnostic ICD codes from insurance
claims data rather than medical record review, which may have caused misclassification bias.
To improve diagnostic validity, IDA cases were identified through diagnostic ICD codes and
procedure codes and TB cases through both diagnostic ICD codes and prescription codes. In
Taiwan, diagnosis of anemia is primarily based on complete blood count; when doctors deter-
mine that the complete blood count reveals microcytic anemia, they order further laboratory
testing on serum iron, ferritin, and TIBA. IDA diagnoses are made when serum ferritin levels
are low. In addition, diagnosis of TB in Taiwan is based on culture and image findings but
sometimes through tissue biopsy and pathological findings. Second, data on alcohol consump-
tion, smoking, homosexual or bisexual behaviors, malnutrition, socioeconomic status, body
mass index, and severity of iron deficiency are unavailable in the NHIRD and are all potential
confounding factors of TB. Consequently, we could not adjust for these variables and conduct
related analysis. To partially address this, we used COPD as a proxy variable for cigarette
smoking, similar to some other studies [29–31]. Third, our longitudinal follow-up study dem-
onstrated an association but not a causal relationship. Moreover, we could not determine
whether the etiology, severity, and duration of IDA were related to the development of TB.
Further studies concerning whether IDA severity is related to TB infection rate should be con-
ducted. Finally, IDA is more common in women; thus, our sample is not representative of the
global TB population; moreover, most Taiwanese people have Chinese ethnicity, and our find-
ings therefore may not be generalizable to other racial groups. Our results, therefore, should be
cautiously interpreted.

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 9 / 11

 Association of iron deficiency anemia with TB

Conclusion
This 12-year nationwide population-based cohort study determined that patients with newly
diagnosed IDA had increased incidence of subsequent TB, regardless of gender and age. Age
had effects on both the strength and duration of the TB association. Future studies are required
to explore the mechanisms underlying these associations. Clinicians are suggested to be aware
of the higher TB risk of patients with new IDA diagnosis and to provide appropriate monitor-
ing of high-risk groups.

Acknowledgments
Acknowledgments to Wallace Academic Editing company for their English editing.

Author Contributions
Conceptualization: Kuo-An Chu, Chun-Hsiang Hsu, Mei-Chen Lin, Yi-Hsin Chu, Yao-Min
Hung, James Cheng-Chung Wei.
Data curation: Mei-Chen Lin.
Formal analysis: Kuo-An Chu, Chun-Hsiang Hsu, Mei-Chen Lin, Yi-Hsin Chu, Yao-Min
Hung, James Cheng-Chung Wei.
Funding acquisition: Mei-Chen Lin.
Investigation: Kuo-An Chu, Yao-Min Hung, James Cheng-Chung Wei.
Writing – original draft: Kuo-An Chu, Chun-Hsiang Hsu, Yao-Min Hung.
Writing – review & editing: Yao-Min Hung, James Cheng-Chung Wei.

References
1. Wang SH, Chung CH, Huang TW, Tsai WC, Peng CK, Huang KL, et al. Bidirectional association
between tuberculosis and sarcoidosis. Respirology. 2019 Feb 5. https://doi.org/10.1111/resp.13482
PMID: 30722101
2. Davies PD. The world-wide increase in tuberculosis: how demographic changes, HIV infection and
increasing numbers in poverty are increasing tuberculosis. Ann Med. 2003; 35:235–43. PMID:
12846265
3. World Health Organization. Global tuberculosis report 2013.
4. World Health Organization. Tuberculosis. Available online:http://www.who.int/mediacentre/factsheets/
fs104/en/index.html. (accessed on 17 Jan 2019).
5. Su VY, Su WJ, Yen YF, Pan SW, Chuang PH, Feng JY, et al. Statin use is associated with a lower risk
of TB. Chest 2017; 152: 598–606 https://doi.org/10.1016/j.chest.2017.04.170 PMID: 28479115
6. GBD Tuberculosis Collaborators. The global burden of tuberculosis: results from the Global Burden of
Disease Study 2015. Lancet Infect Dis. 2018 Mar; 18(3):261–84 https://doi.org/10.1016/S1473-3099
(17)30703-X PMID: 29223583
7. Jeon C Y, Murray M B. Diabetes mellitus increases the risk of active tuberculosis: a systematic review
of 13 observational studies. PLOS Med 2008; 5: e152. https://doi.org/10.1371/journal.pmed.0050152
PMID: 18630984
8. Min J, Kwon SK, Jeong HW, Han JH, Kim YJ, Kang M, et al. End-stage Renal Disease and Risk of
Active Tuberculosis: a Nationwide Population-Based Cohort Study. J Korean Med Sci. 2018 Dec 13; 33
(53):e341 https://doi.org/10.3346/jkms.2018.33.e341 PMID: 30595682
9. Lin H H, Ezzati M, Murray M. Tobacco smoke, indoor air pollution and tuberculosis: a systematic review
and meta-Analysis. PLOS Med 2007; 4: e20 https://doi.org/10.1371/journal.pmed.0040020 PMID:
17227135
10. Lonnroth K, Williams B G, Stadlin S, Jaramillo E, Dye C. Alcohol use as a risk factor for tuberculosis—a
systematic review. BMC Public Health 2008; 8: 289 https://doi.org/10.1186/1471-2458-8-289 PMID:
18702821

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 10 / 11

 Association of iron deficiency anemia with TB

11. Lonnroth K, Williams B G, Cegielski P, Dye C. A consistent loglinear relationship between tuberculosis
incidence and body mass index. Int J Epidemiol 2010; 39: 149–55 https://doi.org/10.1093/ije/dyp308
PMID: 19820104
12. Barboza C E, Winter D H, Seiscento M, Santos Ude P, Terra Filho M. Tuberculosis and silicosis: epide-
miology, diagnosis and chemoprophylaxis. J Bras Pneumol 2008; 34: 959–66 https://doi.org/10.1590/
s1806-37132008001100012 PMID: 19099104
13. Getahun H, Gunneberg C, Granich R, Nunn P. HIV infection associated tuberculosis: the epidemiology
and the response. Clin Infect Dis 2010; 50 (Suppl 3): S201–S207
14. Lai SW, Wang IK, Lin CL, Chen HJ, Liao KF. Splenectomy correlates with increased risk of pulmonary
tuberculosis: a case–control study in Taiwan. Clin Microbiol Infect. 2014 Aug; 20(8):764–7 https://doi.
org/10.1111/1469-0691.12516 PMID: 24372744
15. Cheng KC, Liao KF, Lin CL, Lai SW. Gastrectomy correlates with increased risk of pulmonary tuberculo-
sis: A population-based cohort study in Taiwan. Medicine (Baltimore). 2018 Jul; 97(27):e11388.
16. Steiger Z, Nickel WO, Shannon GJ, Nedwicki EG, Higgins RF. Pulmonary tuberculosis after gastric
resection. Am J Surg. 1976 Jun; 131(6):668–71. https://doi.org/10.1016/0002-9610(76)90174-4 PMID:
937642
17. Maggini S, Pierre A, Calder PC. Immune Function and Micronutrient Requirements Change over the
Life Course. Nutrients. 2018 Oct 17; 10(10). pii: E1531. https://doi.org/10.3390/nu10101531 PMID:
30336639
18. Camaschella C. Iron-deficiency anemia. N. Engl. J. Med. 2015; 372:1832–43. https://doi.org/10.1056/
NEJMra1401038 PMID: 25946282
19. DeLoughery T.G. Microcytic anemia. N. Engl. J. Med. 2014; 371:1324–31. https://doi.org/10.1056/
NEJMra1215361 PMID: 25271605
20. Ganz T. Iron and infection. Int J Hematol. (2018) Jan; 107(1):7–15. https://doi.org/10.1007/s12185-017-
2366-2 PMID: 29147843
21. Kumar V, Choudhry VP. Iron deficiency and infection. Indian J Pediatr. 2010 Jul; 77(7):789–93 https://
doi.org/10.1007/s12098-010-0120-3 PMID: 20589461
22. Tansarli GS, Karageorgopoulos DE, Kapaskelis A, Gkegkes I, Falagas ME. Iron deficiency and suscep-
tibility to infections: evaluation of the clinical evidence. Eur J Clin Microbiol Infect Dis. 2013 Oct; 32
(10):1253–8. https://doi.org/10.1007/s10096-013-1877-x PMID: 23595586
23. Cherayil BJ. Iron and immunity: immunological consequences of iron deficiency and overload. Arch
Immunol Ther Exp (Warsz). 2010 Dec; 58(6):407–15.
24. Oppenheimer SJ. Iron and its relation to immunity and infectious disease. J Nutr.2001; 131: 616S–
635S. https://doi.org/10.1093/jn/131.2.616S PMID: 11160594
25. Macdougall IG, Anderson R, McNab GM, Katz J. The immune response in iron-deficient children:
impaired cellular defense mechanisms with altered humoral components. J Pediatr. 1975; 86: 833–43.
https://doi.org/10.1016/s0022-3476(75)80211-3 PMID: 1127523
26. Laudisio A, Marzetti E, Pagano F, Bernabei R, Zuccalà G. Haemoglobin levels are associated with
bone mineral density in the elderly: A population-based study. Clin. Rheumatol. 2009; 28:145–51.
https://doi.org/10.1007/s10067-008-0998-6 PMID: 18766424
27. Korkmaz U, Korkmaz N, Yazici S, Erkan M, Baki A.E, Yazici M. Anemia as a risk factor for low bone min-
eral density in postmenopausal Turkish women. Eur. J. Intern. Med.2012; 23:154–58. https://doi.org/
10.1016/j.ejim.2011.11.009 PMID: 22284246
28. Pan ML, Chen LR, Tsao HM, Chen KH. Iron Deficiency Anemia as a Risk Factor for Osteoporosis in Tai-
wan: A Nationwide Population-Based Study. Nutrients. 2017 Jun 16; 9(6). pii: E616 https://doi.org/10.
3390/nu9060616 PMID: 28621741
29. Chang KH, Chang MY, Muo CH, Wu TN, Chen CY, Kao CH. Increased risk of dementia in patients
exposed to nitrogen dioxide and carbon monoxide: a population-based retrospective cohort study. PloS
one. 2014; 9(8):e103078 https://doi.org/10.1371/journal.pone.0103078 PMID: 25115939
30. Raaschou P, Simard JF, Asker Hagelberg C, Askling J. Rheumatoid arthritis, anti-tumour necrosis fac-
tor treatment, and risk of squamous cell and basal cell skin cancer: cohort study based on nationwide
prospectively recorded data from Sweden. BMJ. 2016; 352:i262 https://doi.org/10.1136/bmj.i262 PMID:
26823527
31. Chang KH, Hsu YC, Hsu CC, Lin CL, Hsu CY, Lee CY, et al. Prolong Exposure of NSAID in Patients
With RA Will Decrease the Risk of Dementia.Medicine (Baltimore). 2016; 95(10):e3056.

PLOS ONE | https://doi.org/10.1371/journal.pone.0221908 August 30, 2019 11 / 11