Risk of Subsequent Dementia in Patients with
Hypertensive Encephalopathy: A Nationwide
Population-Based Study in Taiwan
Wei-Shih Huang
a, eChun-Hung Tseng
a, eCheng-Li Lin
b, dChi-Yu Lin
fFung-Chang Sung
b, dChia-Hung Kao
c, eIntroduction
The term ‘hypertensive encephalopathy’ (HE) was first used in 1928 by Oppenheimer and Fishberg [1] to describe episodes of acute organic brain syndrome correlated with hypertension,
which is characterized by unspecific neurological symptoms including headaches,
visual disturbances, altered mental statuses, and seizures [2] . The term ‘reversible posterior leukoencephalopathy syndrome’ (RPLS), first used by Hinchey et al. [3] in 1996, has also been
used to refer to similar cases because of their indistinguishable reversible clinical pictures with elevated mean arterial pressure and the same radiological feature in computed tomography
or magnetic resonance images as edema involving white matter in the parietooccipital areas [2–4] . Meanwhile, the association of hypertension with RPLS is well documented but hypertension is not always present in such cases; particularly in patients treated with immunosuppressant,
hypertension may be absent [3] . Therefore, the older term ‘HE’ seems indeed
too restrictive for an appropriate naming, and the term ‘RPLS’ was considered misleading by other authors because the involvement of gray matter was also observed. Thus, the term
‘posterior reversible encephalopathy syndrome’ was introduced [5] . The mechanism of these
possible brain-capillary leak syndromes is suggested to be extreme elevation of systemic
blood pressure, which causes the breakdown of the autoregulatory vascular endothelium of
the brain [3] . Although HE is typically reversible, failure to treat the dramatic rise in blood
pressure may promptly lead to fatal consequences. This emerging syndrome requires
accurate
diagnosis and early management. The precise incidence of HE is unknown. However, hypertensive
crises (76% urgencies, 24% emergencies) were reported to be more than one fourth
of all medical emergencies in Athens in 2010 and can result in acute end-organ injuries such as cerebral infarction, acute myocardial infarction (AMI), heart failure, acute renal failure, and
HE [6] .
Dementia is an emerging health problem for elderly people in Taiwan. Several risk factors are associated with dementia, including age, sex, inflammation, genetic factors, comorbidity, environmental factors, and lifestyle [7] . The association of HE with risk of dementia has rarely
been evaluated. In this study, we investigated whether the diagnosis of HE is linked to an increased risk of developing subsequent dementia by using the Taiwanese National Health Insurance Research Database (NHIRD). The database is available to researchers in Taiwan and has been extensively used in epidemiologic studies [8] . The wide coverage of this large, nationwide database allowed us to examine the relationship between HE and the subsequent development of dementia.
Materials and Methods Data Sources
The National Health Insurance system is a government-operated, single-payer health insurance program, which was launched in 1995 and covered approximately 99% of the 23.74 million people in Taiwan
by 2009 [9] . With authorization from the Department of Health, the National Health Research Institutes (NHRI) is responsible for managing the claims data from all health providers and available information on all insurants. The NHRI established several databases for academic and administrative uses, providing information
on the basic patient demographic status, diagnostic codes, medical institutions, outpatient and inpatient orders, prescriptions, and expenditures for care. Data files are linked with scrambled patient identification numbers to protect the privacy of the patients. We used the nationwide population-based data for the period from 1996 to 2010 released by the NHRI. The International Classifications of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes were used to define diseases in the claims data. We confirm that all data were de-identified and analyzed anonymously. In addition, this study was approved by the Ethics Review Board at China Medical University (CMU-REC-101–012). Several Taiwanese studies had demonstrated
the high accuracy and validity of ICD-9 diagnosis in NHIRD [10, 11] . Study Subjects
Figure 1 shows the selection process of the subjects in the 2 study cohorts. From the claims data for
inpatients from 1997 to 2010, we identified patients with the diagnosis of hypertension (ICD-9-CM codes 401–405). Patients with newly diagnosed HE (ICD-9-CM code 437.2) 40 years of age and older were selected as the study cohort. The date of diagnosis was defined as the index date for initiating follow-up person-years measurement. The comparison cohort consisted of patients randomly selected from the rest of the
hypertensive
patients without an HE history. For each HE case, we randomly selected 4 subjects from the comparison cohort who were frequency-matching with sex, age (every 5-year span), and index date. Patients with any history of dementia and stroke before the index date were excluded.
Outcome Measures
The follow-up person-years were measured for each patient from the index date to the date the patient was diagnosed with dementia (ICD-9-CM codes 2900–2904), the date he/she was censored because of loss to follow-up, died, withdrew from the insurance system or until the end of 2010. A baseline comorbidity history including diabetes (ICD-9-CM code 250), AMI (ICD-9-CM code 410), head injury (ICD-9-CM codes 850–854, 959.01), and depression (ICD-9-CM codes 2962–2963, 300.4, 311) was determined from inpatient claims data for each participant.
Statistical Analysis
Demographic characteristics and comorbidity prevalence were compared between the study and
comparison cohorts. The variations were examined using the test for categorical variables and the test for continuous variables. The dementia incidence rates were estimated according to demographic status and comorbidity prevalence and compared between the 2 cohorts. Death occurring prior to dementia was considered a competing risk event in the analysis. The univariable and multivariable modified Cox proportional
hazards regression model was used to assess the risk of developing dementia associated with HE
compared with the non-HE cohort. The multivariable model was simultaneously adjusted for age, sex, and the comorbidities of diabetes, AMI, head injury, and depression. The related hazard ratio (HR) and 95%
confidence
interval (CI) were estimated in the modified Cox model. The follow-up years were further stratified into quartiles to observe the change of dementia hazard. The data analysis also evaluated the interaction between HE and head injury. All analyses were performed using the SAS statistical package (version 9.1; SAS Institute Inc., Cary, N.C., USA). A 2-tailed p value <0.05 was considered statistically significant.
Results
We identified 5,504 patients with HE for the study cohort and selected 22,016 patients
from the patients with hypertension for the comparison cohort, with mean ages of 68.2
(SD = 11.9) and 68.1 (SD = 11.9) years, respectively ( table 1 ). Comorbidities, particularly
diabetes, AMI, head injury, and depression, were prevalent in the study cohort. The mean
follow-up period for identifying dementia was shorter for the HE cohort than for the non-HE
cohort (5.56 ± 3.97 vs. 7.63 ± 3.70 years). The dementia incidence was 1.45-fold (95% CI =
1.27–1.66) higher in the study cohort than in the comparison cohort (8.92 vs. 5.70 per 1,000
person-years; table 2 ). In the study cohort, the dementia incidence was higher for males, advanced age, and patients with comorbidities, excluding AMI. An age-specific analysis indicated
that the crude HR between the study cohort and the comparison cohort was higher for younger than for elderly patients (crude HR = 2.33, 95% CI = 1.47–3.68 vs. crude HR = 1.43, 95% CI = 1.24–1.64). However, the adjusted HR was 7.01 (95% CI = 6.80–10.9) for the elderly
patients compared with the participants aged 40–64 years. The dementia incidence also increased for those with a comorbid head injury. Table 3 shows the dementia incidence associated
with the interaction of HE, diabetes, and head injury. The dementia incidence was
much higher in patients with HE, head injury, or diabetes. Compared to those without HE, diabetes, or head injury, HE patients with head injury exhibited an adjusted HR of 3.58 (95%
CI = 2.15–5.96), followed by those with simultaneous diabetes and head injury, exhibiting an adjusted HR of 2.09 (95% CI = 1.18–3.71). Figure 2 shows the association between HE and the risk of developing dementia by the quartiles of the follow-up years. The risk of dementia was highest in the first quartile of ≤ 4.10 years (HR = 2.07, 95% CI = 1.70–2.52). The HR declined thereafter to 0.38 (95% CI = 0.16–0.88) in the fourth quartile.
Discussion
According to our review of relevant research, this study is the first attempt at investigating the risk of dementia in patients with HE after adjusting for comorbid medical disorders by using a nationwide population-based data set. Our study demonstrated that the likelihood of developing dementia is 1.58-fold (95% CI = 1.37–1.83) greater in hypertensive patients with HE than in those without HE. Furthermore, we observed that hypertensive patients with
HE and the comorbidities of both head injury and diabetes exhibited a substantially higher risk (HR = 3.40, 95% CI = 1.92–6.03) of dementia than those without HE and both
comorbidities.
The overall demographic-specific dementia incidence after the stratified analyses of sex and age was significantly higher in the cohort with HE than in the comparison cohort.
Moreover, the relative dementia incidence was highest in the first quartile of ≤ 4.10 years (HR = 2.15, 95% CI = 1.78–2.59) and gradually decreased in the later quartiles of the followup
years. This finding further confirms that the risk of dementia in patients with HE is truly increased compared to the non-HE cohort.
Several epidemiological studies have demonstrated that the prevalence of dementia in
Taiwan is between 1.7 and 4.3% in adults aged ≥ 65 years [12] . All of these studies have
observed that the prevalence of dementia increases with age and is higher in women than in
men. Dementia is categorized into subtypes according to its causes. Alzheimer’s disease accounts for approximately half of the affected population in Taiwan, followed by vascular dementia (20–25%) and mixed dementia (5–10%) [13] . According to data from the Association
of Dementia in Taiwan, dementia affects more than 160,000 subjects. The annual dementia incidence among subjects aged ≥ 65 years in Taiwan is 1.3%, with the incidence risk increasing with age from 0.77% for people aged 65–74 years to 6.19% for those aged ≥ 85 years [14] . Dementia is an emerging health problem for elderly people in Taiwan. Factors associated with dementia include age, sex, inflammation, genetic factors, comorbidity, environmental
factors, and lifestyle. Protective factors include a high education level, moderate alcohol consumption, use of hormone replacement therapy for women, use of anti- inflammatory
drugs, and diet [7] . The association between HE and the risk of dementia has rarely been evaluated. Changes in the degree of hypertension can generate autoregulatory
dysfunction of the cerebral blood flow [15] . Recent reports have asserted that HE occurs in 15–20% of the patients in whom malignant hypertension is present [16, 17] . HE is an acute organic brain syndrome caused by disrupted autoregulation of the cerebral blood flow.
Vasogenic
edema is related to hypertensive cerebrovascular endothelial dysfunction or disruption of the blood-brain barrier with increased permeability [4, 15, 18] . The disruption is generally
the most accepted basis for HE [19] . Widespread cerebral edema can occur with rapidly progressive hypertension. Atherosclerosis, arterial and arteriolar fibrinoid necrosis, and microaneurysms may be observed. HE may therefore cause vascular dementia and can be associated with subcortical arterial and arteriolar leukoencephalopathy, leukoaraiosis, and Binswanger’s disease [20] .
The use of a nationwide population-based data set that provided a sufficient sample size and statistical power to explore the link between HE and dementia is a particular strength of this study. In addition, the patients in our study displayed a wide range of demographic characteristics,
which allowed us to perform stratified analyses according to sex, age, and comorbidities.
Nevertheless, insufficiencies in our study should be addressed. First, we cannot
completely exclude study participant ‘misclassification’. A patient with HE might not have
sought medical advice and thus been misclassified as having only hypertension, leading to
his/her inclusion in the comparison cohort. This probability was extremely low because few
patients would tolerate acute HE symptoms without medical intervention. Moreover, our
patients could seek medical advice easily because of the high accessibility of medical services
in Taiwan. The diagnosis of HE in Taiwan is based on conventional clinical syndromes described by Oppenheimer and Fishberg [1] and findings from brain CTs or MRI. The accuracy
of HE coding is supposed to be nearly 100% because brain CTs are always available and performed in every Taiwanese hospital treating HE, which is an emergent condition [21–23] . Delay in the treatment of HE may in fact result in death or irreversible neurological sequelae
[24] . Second, because there are no specific ICD-9 numbers for RPLS and posterior reversibleencephalopathy syndrome and they both have similar clinical presentations and radiological features to HE [25–28] , their ICD-9 number was therefore coded as HE. Third, additional, theoretically relevant confounding variables such as smoking, lifestyle, and drugs could not be included in our analysis because they were not included in our data set. Further study is required to clarify the effects of these factors. Fourth, a possible confounder is that patients with HE may differ in terms of socioeconomic class or education and hence might be less compliant with medications or less likely to seek medical attentions.
Conclusion
