Original Article
Thalassemia and risk of dementia: A nationwide population- based
retrospective cohort study☆
Yu-Guang Chen a, Te-Yu Lin b, Hsuan-Ju Chenc,d, Ming-Shen Dai a, Ching-Liang Ho a, Chia-Hung Kao e,f,⁎
1. Introduction
Thalassemia is a common hereditary hemoglobinopathy resulting from signifcantly impaired synthesis of one or more globin subunits of normal hemoglobins. Clinical classifcation of thalassemia is based on a mutation or a defcient amount of globin numbers. The data bank of the World Health Organization has indicated a recent prevalence of hemoglobinopathies reported in approximately 5% of the global population [1]. Diverse clinical characteristics varying from asymptomatic to
transfusion-dependent severe ineffective erythropoiesis have been
documented in thalassemia patients, typically resulting fromsevere hemoglobinopathies.
Mainstay pathophysiology in thalassemia patients
includes ineffective erythropoiesis and peripheral hemolytic anemia caused by membrane destruction [2]. Several chronic complications and subsequent end-organ damage have been observed with disease severity, repeated blood transfusions, and impaired gastrointestinal iron absorption contributing to abnormal iron deposition [3]. Because of advances in prenatal screening and diagnostic methods, the severe genotypes of thalassemia have declined gradually, and most cases are silent carriers or have a mildly lower hemoglobin value without any symptoms, and are transfusion-independent or transfusion-naïve population [4].
Based on DSM-5 classifcation, neurocognitive disorders included delirium, mild cognitive impairment and dementia. Dementia is a major neurocognitive disorder characterized by impaired function of learning and memory, executive function, perceptual motor, complex attention, social cognition and language that can result in severe disabilities
in daily activities, leading to a great health care burden in the general population worldwide [5].
Several defned risk factors included
aging, presence of apolipoprotein E4 allele, socioeconomic status, and vascular factors such as hypertension, diabetes, head injury, and depression [6]. Based on recent studies, anemia is another strong risk factor for
dementia pathogenesis [7,8]. A large systematic review also revealed that anemia is signifcantly associated with cognitive performance [9].
A population-based cohort study has also shown a positive correlation of iron defciency anemia with dementia [10]. However, these studies have focused on acquired anemia, whereas few studies have focused on hereditary anemia. Previous studies with small sample sizes have demonstrated a trend of developing cognitive dysfunction in severe types of thalassemia and the hemoglobin mutations associated with β thalassemia minor [11–13]. However, studies elucidating the relationship between thalassemia and long-term dementia incidence are rare.
This nationwide population-based cohort study clarifed the association between thalassemia and dementia in Taiwan population.
2. Methods 2.1. Data source
The National Health Insurance (NHI) Program implemented on March 1, 1995 covers more than 99% of the 23.74 million population of Taiwan. The current study was a retrospective cohort study that
used the Longitudinal Health Insurance Database 2000 (LHID 2000), released by the National Health Research Institutes (NHRI). The NHRI
established a National Health Insurance Research Database (NHIRD) to record all benefciary medical services, including inpatient and outpatient demographics, primary and secondary diagnoses, procedures, prescriptions, and medical expenditures. The LHID 2000 includes one
million insurants randomly selected from the 2000 Registry of NHI benefciaries, which contains all medical records of each insurant from1996
to 2011. No signifcant difference exists in sex, age, or health care costs between cohorts in LHID 2000 and all insurance enrollees, as reported by theNHRI in Taiwan [14]. Diseasewas identifed based on the International Classifcation of Diseases, 9th Revision (ICD-9 codes).
2.2. Sampled participants
We conducted a cohort analysis to determine the association between thalassemia (ICD-9-CM code 282.4) and dementia (ICD-9-CM
codes 290.0–290.4, 294.1, and 331.0–331.2). Patients aged more than 20 yearswhowere identifedwith newly diagnosed thalassemia during
January 1, 2000 and December 31, 2004, and who had no history of dementia, low B12 (ICD-9-CM codes 281.1 and 281.2), folic acid anemia
(ICD-9-CM codes 280.0, 280.1, 280.8, and 280.9), or receiving transfusion (ICD-9 code for procedure 990), were included as the thalassemia
cohort. The index date for each thalassemia case was the date of thalassemia diagnosis. To increase statistical power, for each patientwith thalassemia, 4 study subjects without history of anemia (ICD-9-CM codes
280–285), dementia, or receiving transfusion were selected from the LHID 2000 as the no thalassemia cohort, and frequency-matched for sex, age (per 1 year) and the year of thalassemia diagnosis. Overall, 871 patientswere included as the thalassemia cohort and 3484 patients as the no thalassemia cohort.
2.3. Outcome and comorbidities
Person-years of follow-up were calculated for each patient until dementia diagnosis, death, withdrawal from the insurance system, or until the end of 2011. Baseline comorbidities associated with dementia were also analyzed. These included diabetes (ICD-9-CM code 250), hypertension (ICD-9-CM codes 401–405), coronary artery disease (CAD, ICD-9-CM codes 410–414), head injury (ICD-9-CM codes 850–854 and 959.01), depression (ICD-9-CM codes 296.2, 296.3, 300.4, and 311), chronic kidney disease (CKD, ICD-9-CM code 585), stroke (ICD-9-CM codes 430–438) and substance-related disorder (ICD-9-CMcodes 291–299 and 303–305).We also used the transfusion as the severity marker of thalassemia to evaluate thalassemia disease status. Transfusion in the follow-up period served as a clinical relativeseverity manifestation for thalassemia patients. In addition, patients
had received a diagnosis of dementia on two separate occasions, and the diagnoses wasmade by an Neurologist clinical evaluation, serial imaging and laboratory investigations during the period between 2001
and 2011.
2.4. Statistical analysis
The distributions of age and histories of comorbidities were compared between the thalassemia and no thalassemia cohorts. The differences of categorical variables were analyzed using the chi-square test, and differences of continuous variables were estimated using t-tests.
The gender-, age- and comorbidity-specifc incidence density rates of dementia were calculated in follow-up period until the end of 2011 or the date of the frst diagnosis of dementia, death, and loss to followup.
We used the Kaplan–Meiermethod to compare the cumulative incidence rates of dementia in cohort of participants without thalassemia,
with thalassemia and with transfusion, and with thalassemia and without transfusion, and used the log-rank test to examine the differences
among the survival curves. The Cox proportional hazard models were used to determine the association of thalassemia and dementia. Hazard ratios (HRs) and their respective 95% confdence intervals (CIs) after adjusting for gender, age, insured amount, urbanization, diabetes,
hypertension, CAD, head injury, depression, CKD, stroke and substancerelated disorder were reported. All data analyses were performed using
SAS 9.3 statistical package (SAS Institute Inc., NC, USA), with p b 0.05 in 2-tailed tests considered signifcant.
2.5. Data availability statement
All data and relatedmetadatawere deposited in an appropriate public repository. The data on the study population thatwere obtained from the NHIRD (http://w3.nhri.org.tw/nhird//date_01.html) aremaintained in the NHIRD (http://nhird.nhri.org.tw/). The NHRI is a nonproft foundation established by the government.
2.6. Ethics statement
The NHIRD encrypts patient's personal information to protect privacy and provides researchers with anonymous identifcation numbers associated with relevant claims information, including sex, date of birth, medical services received, and prescriptions. Patient consent is
not required to access the NHIRD. This studywas approved by the Institutional Review Board (IRB) of China Medical University (CMU-REC-
101-012). The IRB specifcally waived the consent requirement.
3. Results
In total, 871 patients with thalassemia and 3484 subjects without
thalassemia were enrolled in our study, with similar age and sex distributions in both. Among the study patients, 8.96%were 65 years of age or
older (Table 1). Themean age in the thalassemia and no thalassemia cohorts was 38.91 years [standard deviation (SD) = 15.45 years] and
38.92 years (SD = 15.46 years), respectively. In addition, we observed much higher urbanization level in non-thalassemia patient than thalassemia patients (p = 0.004) However, we didn't fnd any signifcant
difference in insured amount between cohort with and without thalassemia (p=0.10). The thalassemia cohort had a higher prevalence of diabetes (7.12% vs 4.91%), CAD (11.60% vs 7.29%), depression (5.05% vs 2.15%), and CKD (1.95% vs 0.43%) than did the no thalassemia cohort.
The mean follow-up years were 8.84 years (SD = 2.20 years) and 8.86 years (SD=2.11 years) for the thalassemia cohort and the no thalassemia
cohort, respectively (data not shown).
Fig. 1 shows a higher cumulative incidence of dementia in the thalassemia
cohort than that in the no thalassemia cohort. And in thalassemia group, there was higher cumulative incidence of dementia in patients with transfusion than those without transfusion. The overall incidence density rate of dementia was higher in the thalassemia cohort than in the no thalassemia cohort (2.86 and 1.46 per 1000 person-y, respectively), with an adjusted HR of 1.88 (95% CI=1.10–3.21) (Table 2). Stratifed
by gender, the incidence rates of dementia inwomen andmen with thalassemiawere 3.36 and 2.16 per 1000 person-years, respectively. In addition,
women displayed 3.22-fold (95% CI = 1.66–6.24) higher risk of developing dementia than the no thalassemia cohort. Furthermore,
relative to the no thalassemia cohort aged over 65 years, thalassemia patients aged over 65 yearswere 2.24-fold more likely to develop dementia
(95% CI=1.17–4.29). Among the patients without diabetes, depression, stroke, or substance-related disorder those with thalassemia were more likely to develop dementia compared with those without thalassemia (adjusted HR = 2.01, 95% CI = 1.07–3.81; 2.54, 1.46–4.12; 1.86, 1.06–3.24; and 1.98, 1.11–3.55, respectively). Among the patients with hypertension, those with thalassemia were also more likely to have dementia compared with those without thalassemia (adjusted HR =
2.20, 95% CI =1.13–4.29).
We also observed a higher risk of dementia for patients with thalassemia and DM, hypertension, CAD, head injury, CKD and substancerelated disorder compared with patients with no thalassemia and no counterpart comorbidity (Table 3). Overall, the HRs of dementia increased, from 1.69 (95% CI = 0.93–3.07) for those who had not
undergone transfusion to 2.72 (95% CI = 1.09–6.78) for those experienced transfusion compared with the no thalassemia cohort (p for
trend b 0.01) (Table 4).
4. Discussion
This is the frst nationwide population-based cohort study involving
an extremely large database that was adjusted for numerous classic dementia risk factors. The fndings suggest thalassemia is a potential risk
factor for developing dementia. Our current study showed that patients with thalassemia have a 1.88-fold risk for developing dementia compared with patients without thalassemia, after we adjusted for age,
gender, insured amount, urbanization and medical comorbidities. In subgroup analysis, there was increased risk to 2.72-fold for developing
dementia in thalassemia patients had experienced transfusion. There
are several strengths to this novel study. First, theNationwideHealth Insurance database in Taiwan includes data for almost all persons, an unparalleled resource among low-, middle-income countries, and one that
is extremely rare even in the high income countries, outside Scandinavia.
Second, in previous study, that have reported worse cognitive function in persons (often children) with relatively small sample size of
thalassemia patients, but our current paper based on over 800 persons represents valuable data. Third, thalassemia is a fairly common condition in certain parts of the world (Mediterranean, South and East Asia) and this paper represents a very laudable effort at identifying novel risk factors for dementia.
A cross-sectional study of elderly patients recognized several comorbidities from large-scale, population-based studies. These included hypertension, diabetes, and other comorbidities such as congestive heart
failure, cerebrovascular disease, and cardiac arrhythmia that have been signifcantly associated with dementia [15]. Anemia is a recognized risk factor as well as above risk factors to develop dementia. In a
large scale cohort study in elderly patients, they found anemia is an independent risk factor after adjusting for detail characteristics such as
demographics, APOE ε4, baselineModifed Mini-Mental State score, comorbidities, and renal function [16]. Another cohort study of 558
community-dwelling olderwomen demonstrated that age-related neurodegenerative effects play crucial roles in cognitive and physical decline.
A relatively lower hemoglobin value and increased decline of cognitive and physical functions were also elucidated [17].
Thalassemia is a hereditary anemic disease that contains various mutations such as deletion of α- or β-chains of hemoglobin, subsequently resulting in 2 main pathogeneses of thalassemia-related
anemia, ineffective erythropoiesis and hemolytic anemia [18,19].
Although the NHRI database in Taiwan using the ICD-9 CM code
provides no information on the underlying genotype and associated hemoglobinopathies, wewere able to roughly estimate the genotype population
based on the previous epidemiology results of thalassemia in
Taiwan. The prevalence rate of α and β thalassemia was approximately 3–5% and 1–3%, respectively. And most cases involved were hereditary
α- and β-thalassemia traits/minor in comparison to relatively less cases of thalassemia major. Those patients of α- and β-thalassemia
traits/minor exhibited asymptomatic or minimal symptoms with transfusion-naïve characteristics [20–22]. We could clarify severity of clinical characteristics according to the transfusion ICD-9 CM code to separate our patients into transfusion-naïve thalassemia populations such as silent carrier or trait, and thalassemia intermedia/major types.
Generally, the latter usually had more severe clinical manifestations and relatively lower serum hemoglobin caused by ineffective erythropoiesis and hemolytic anemia. The possible genotype classifcation in
these thalassemia patients that required transfusion was thalassemia intermedia or major type. They could require transfusion to maintain daily activity, particularly in co-existing with more comorbidities or complications [23]. Our results showed that there was mild increase in incidence rate in thalassemia patients who did not receive blood transfusion despite there was no statistical signifcance compared with the
general population. However, thalassemia patients who underwent transfusion had a signifcant increase in incidence for developing dementia.
The possible explanation of our study that could be considered is
that thalassemia patientswho required transfusion frequently had relatively severe clinicalmanifestation such as hemolytic anemia or ineffective hematopoiesis. Vascular endothelial cell injury with endovascular infammatory processes caused by hemolytic anemia, ischemic stroke or poor tissue perfusion could lead to Alzheimer's disease or vascular dementia [24–26]. In addition, a profound hemostatic change with a hypercoagulable state has also been recognized in patients with β- thalassemia major, β-thalassemia intermedia, and α-thalassemia (hemoglobin H disease) [27]. Some studies have also primarily reported cerebral thrombotic events in patients with β-thalassemia, and other thromboembolic events including deep venous thrombosis, pulmonary embolism, silent cerebral infarction, or recurrent arterial occlusion observed in serial studies [28–30]. Therefore, these effects, including vascular injury, anemia, and a hypercoagulable state related to silent stroke or small vessel disease may increase the risks of developing long-term dementia in thalassemia patients.
In current studies, we found that the incidence rate of dementia increased in thalassemia patients with chronic medical comorbidities compared with patients with only chronic comorbidities, but
without thalassemia. A relatively high incidence rate was recognized
in thalassemia patients experiencing head injury, compared with the nonthalassemia population. Traumatic head injury is a well-known risk factor for developing dementia [31,32]. However, these may be explained that there were too small sample size (only two cases) in persons
with both thalassemia and head injury. The role of head injury
and thalassemia should be investigated based on further detail studies.
We also found an increased risk in female thalassemia patients for developing dementia. This is possibly because female patients sometimes
experience combined thalassemia and iron defciency anemia (IDA).
However, IDA is typically neglected because both belong to microcytic
anemia, and eventually these patients present with relatively lower hemoglobin value [33]. This effect could be the possible etiology of the
high risk in female thalassemia patients for developing dementia in the long-term follow-up. Similarly for the higher risk among women compared to men, it seems rash to speculate that this is due to coexistent occult iron-defciency anemia as many other possible mechanisms
exist for a difference in risk by gender. This study has several limitations. First, the NHIRD does not contain
detailed information regarding patients' current use of medications, smoking habits, and obesity, all of which might be confounding factors in this study. Second, the thalassemia genotype is important, but is not provided by the NHIRD database. The various genotype classifcations could also lead to heterogeneous clinical manifestations. Suffcient clinical data does not exist based on the ICD-9 CMcode alone. The details of individual differences should be elucidated according to studies that are more complete. Although we used the transfusion ICD-9 CM code
to distinguish thalassemia major group, we did not obtain certain detailed information of thalassemia patients including hemoglobin level,
clinical manifestations, and chronic complications. In thalassemia patients without transfusion, we can't distinguish thalassemia trait from
non-transfusion dependent thalassemia patients such as HbH disease/
thalassemia intermedia. This group presented relatively severe clinical presentation and could be a confounding factor in our current study.
Third, the evidence derived from a retrospective cohort study is typical of lower quality in statistics than that derived from prospective cohort study because of potential biases related to adjustments for confounding variables. Fourth, therewere relativelymore comorbidities in thalassemia groups than in non-thalassemia patients. It is likely that
persons who receive medical attention for any cause are more likely to be diagnosed earlier as having dementia, especially since the database has shown that there may be an under-diagnosis of psychogeriatric and psychiatric diseases in Taiwan as a whole. This effect appears to show an increased risk of dementia among persons more likely to receive frequent medical care. Fifth, although mean follow-up time of 8.84 and 8.86 years seems like a reasonable time at the frst glance, considering the mean age of the study population is 38.9 years, and the dementia is mostly seen after 65 years of
age, most of the study population did not reach the age to develop dementia.
To test the hypothesis of this study much longer follow-up is necessary.
The main contribution of our study is its use of a nationwide
population-based database that contains data on a high number of thalassemia cases. The results indicated that thalassemia is a potential risk
factor for developing dementia and thalassemia patients that experienced transfusion exhibit a 2.72-fold risk for developing dementia, compared with the general population. And there was increased risk in
thalassemia patients with chronic comorbidities. Although there was no signifcant increased risk for developing dementia in most thalassemia patients (transfusion-naïve thalassemia patients) statistically, we
had observed a trend of increased risk for dementia in those patients.
More detail studies with longitudinal follow-up should be conducted to investigate clear association between this population and dementia.
We recommend that physicians consider the possibility of dementia when they observe patients with thalassemia, particularly in the presence of dementia-related comorbidities, some of which may be potentially treatable to management.
