Acupuncture Decreases the Incidence of Poststroke Dementia: A Taiwan Nationwide Retrospective Cohort Study

Stroke has been recognized as one of the major causes of adult disability worldwide, and post-stroke dementia may affect up to one-third of stroke survivors. This condition could be prevented if we could recognize and treat it earlier. Acupuncture as a complementary treatment for stroke has been shown to be benecial for subsequent post-stroke rehabilitation. Our study investigated if acupuncture served additional advantages to decreasing the incidence of post-stroke dementia. We provide a retrospective cohort study from Taiwan National Health Insurance Research Database. This study compared the incidence of post-stroke dementia between cohorts with and without acupuncture treatment by calculating adjusted hazard ratios (HRs) and 95% condence intervals (CIs) of acupuncture associated with post-stroke dementia using Cox proportional hazard regressions. The study group was composed of 9,547 patients who received rst hospitalization for the treatment of stroke between 2002 and 2004; of which, 47,735 comparison patients were dened as the control group. The two groups were followed-up until the diagnosis of dementia presented or until the end of 2007. follow-up conducted a retrospective, nationwide population-based cohort study to investigate the possibility of in reducing of post-stroke


Background
Stroke has been recognized as one of the major causes of adult disability worldwide. Survivors of stroke are at increased risk of developing cognitive impairment due to acute tissue damage. Assuredly, one-third of these survivors are found to have a signi cant degree of cognitive impairment within the rst month after the event [1][2][3]. The lifetime risk of developing either stroke or dementia by age of 65 is one in three in men and one in two in women [4]. Post-stroke condition constitutes a well-de ned reason of dementia rst recognized by Sir Thomas Willis in 1672 under the name postapoplectic dementia [2]. This is an etiological classi cation of dementia characterized by cognitive impairment as a result of ischemic or haemorrhagic stroke. Hypoperfusion and thromboembolism events lead to decreased cerebral blood ow, hypoxia, and oxidative stress and trigger in ammatory responses, subsequently damaging the area of brain important for memory, cognition, and behaviour [5]. Although functional recovery develops over the course of 26 weeks after a stroke, and physical impairments tend to improve to a greater or lesser degree, the survivors are often left with disabilities and worsening cognitive impairments [6,7].
Currently there is no de nitive and effective treatment for post-stroke dementia. Treatments with antioxidants, anti-in ammatory agents or agents increasing cerebral perfusion have not provided satisfactory results [8]. Numerous clinical studies and systemic reviews support the therapeutic effects of acupuncture in vascular dementia (VD) and post-stroke rehabilitation [9][10][11]. However, these therapeutic effects remain unclear because of poor methodology and small sample sizes [12,13]. Acupuncture, with many categories such as manual acupuncture, electroacupuncture, laser acupuncture, and acupoint injection, serves as a form of alternative treatment and an essential component of traditional Chinese medicine which can be traced back 3,000 years [9]. It is a relatively safe procedure with few adverse effects. The mechanisms of acupuncture on VD are still uncertain; however, animal-based studies demonstrate the possibility of reducing oxidative stress, attenuating neuronal apoptosis, relieving neuroin ammation, regulating glucose metabolism, modulating neurotransmitters, and improving synaptic plasticity and blood vessel function [14]. Acupuncture has become universally accepted. The 2007 US National Health Interview Survey found 3.1 million adults and 150,000 children had received acupuncture in the previous year, and acupuncture use increased overall between 2002-2007 [15,16].
It is unknown if acupuncture is an adequate therapeutic method to reduce post-stroke dementia. Using the Taiwan National Health Insurance Research Database(NHIRD), we conducted a retrospective, nationwide population-based cohort study to investigate the possibility of acupuncture use in reducing incidence of post-stroke dementia.

Database
We conducted a retrospective, population-based cohort study of patients who are registered in the Longitudinal Health Insurance Database (LHID2000). 98.4% of the population in Taiwan (approximately  22.96 million) has been enrolled in Taiwan's NHIRD, which comprises healthcare data from the medical records of all bene ciaries [18,19]. The LHID2000 is composed of 1 million insured subjects randomly selected from the NHIRD without bias to age and gender. The International Classi cation of Diseases, Ninth Revision, Clinical Modi cation (ICD-9-CM) diagnostic and procedure codes (established according to the World Health Organization criteria) are included in the diagnostic data [17]. Details of the generation, monitoring, and maintenance related to the database are published online by the Taiwan's National Health Research Institutes (http://nhird.nhri.org.tw/) [18,19].

Study sample
Patients with a diagnosis of stroke (ICD-9-CM code 430-437), receiving treatment in the hospital between January 1st, 2002 and December 31st, 2004 were enrolled in this study [18]. A ow chart of this study is shown in Fig. 1. We excluded patients aged younger than 40 to restrict the study to the assessment of the most common aetiology of stroke. In order to con rm that all patients with stroke enrolled in this study were incident cases, only new-onset stroke patients were included. Patients with the diagnosis of stroke or dementia before 2001 were excluded. Furthermore, patients who had the following conditions prior to the initial use of health care facilities were identi ed and excluded from this study: ICD-9-CM code 290.0 (senile dementia, uncomplicated); 290.1 (presenile dementia); 290.2 (senile dementia with delusional or depressive features); 290.3 (senile dementia with delirium); 290.4 (arteriosclerotic dementia); 294.1 (dementia in conditions classi ed elsewhere); 331.0 (Alzheimer's disease, AD); 331.1 (Pick disease); and 331.2 (senile degeneration of the brain). A total of 9,547 subjects who had strokes from 2002 to 2004 were matriculated in this study.
We built up a comparison cohort by extracting randomly from the records of the remaining patients in LHID2000 with identical exclusion criteria. The control group comprised 47,735 selected age-and sexmatched individuals without a previous history of stroke or dementia, such that the ratio of comparison subjects to stroke patients was 5:1.

Main outcome measures
We established the primary endpoint of this study as the occurrence of dementia in both genders during a 3-year period from the individuals' initial use of health care. Then, acupuncture (procedure code: B41, B42, B45, B46, P27041, P33031) data was collected to determine whether it is an effective treatment in reducing the incidence of post-stroke dementia. We also categorized the data for these dementia subtypes; AD (ICD-9-CM code 331.0), VD (ICD-9-CM code 290.4), and unspeci ed dementia (ICD-9-CM codes 290.0 to 290.4, 294.1, and 331.1 to 331.2).

Statistical analysis
Cases and controls were followed from the initial hospitalization date until December 31st in 2007. Pearson's chi-square test was administered to compare the baseline characteristics of the stroke cases receiving acupuncture and non-stroke controls from the NHIRD. Selected comorbidities were included if they occurred in either an inpatient setting or 3 or more ambulatory care claims that had been recorded 1 year before the initial ambulatory care visit. Multivariate Cox proportional hazard regression model was conducted to evaluate the hazard ratios (HRs) with 95% con dence intervals (CIs) for risk of dementia in patients receiving acupuncture after occurrence of stroke. Age, gender, and selected comorbidities were adjusted. Apart from the abovementioned comorbidities, we also adjusted for acupuncture in the regression modelling. A p value < 0.05 was de ned as statistical signi cance in this study. The 5-years disease-free survival rate was evaluated with the Kaplan-Meier log rank test to examine differences between cohorts. A statistical analysis software system (SAS, System for Windows, version 9.2, SAS Institute, Inc., Cary, NC, USA) was used to perform the statistical analyses.

Results
Overall, 9,547 and 47,735 patients were included in the stroke and non-stroke cohorts, respectively. There were no signi cant differences in age or gender between these groups. Geographic characteristics and comorbidities of the patients with stroke and comparison patients are shown in Table 1. The proportion of men was higher than that of women in both cohorts (53.48% vs. 46.52%). The mean age of the patients was 66.02 ± 11.72 years. After matching for age and gender, the comorbidities of diabetes, dyslipidaemia, hypertension, coronary heart disease, heart failure, atrial brillation, peripheral vascular disease, respiratory system, peptic ulcer disease, chronic liver disease, chronic kidney disease, rheumatologic disease, and cancer were more prevalent in the stroke cohort, compared with the nonstroke cohort.  Figure 2 indicates the disease-free survival curves determined using the Kaplan-Meier log rank analysis.  Table 3 shows the relative risk of developing distinct types of dementia after acupuncture treatment in the stroke patient group. The analysis consistently demonstrates that, compared to stroke patients without acupuncture, the study patients were less likely to sustain several subtypes of dementia during the 3-year follow-up period after their initial assessment. Notably, the adjusted HRs for other dementia in patients with acupuncture was 0.49 (95% CI, 0.41-0.59; p < 0.001). Figures 3 and 4 disclose the diseasefree survival curves of the patients with stroke and comparison cohorts who did and did not undergo acupuncture using the Kaplan-Meier method. The crude and adjusted HRs of dementia among patients with stroke and those who have received acupuncture in this study are shown in Table 4. Adjusted HR of dementia for patients with stroke was 4.79 (95% CI 4.44-5.18, p < 0.001) and surprisingly, the result reveals that patients who received acupuncture, whether they have a history of stroke or not, are at lower risk of developing dementia during this 3-year follow-up.

Discussion
The consequence of this retrospective, nationwide, population-based cohort study illustrates signi cant associations between post-stroke dementia and application of acupuncture. The best methods to prevent post-stroke dementia are to impede recurrence of stroke and reduce severity of stroke-related impairment through the most favourable acute treatment and intensive secondary prevention [20][21][22]. Secondary prevention consists of medical interventions and lifestyle modi cation. To date, advanced treatment for vascular risk factors and vascular disease may prevent post-stroke dementia. An observational study unveiled reduced cognitive impairment in patients with appropriate post-stroke vascular risk management, including antiplatelet therapy, antihypertensive drugs and statins, or anticoagulants, as indicated [21]. Furthermore, cognitive function can bene t from therapy for neuropsychiatric symptoms like depression, apathy, and anxiety, as well as cognitive training/stimulation [22]. According to our results, we suggest that early intervention with acupuncture could also decrease the risk of post-stroke dementia.
Acupuncture as a traditional Chinese healing technique treats disorders by inserting needles into speci c acupoints, and it is generally considered safe when performed correctly. The use of acupuncture treatment for 43 diseases has been recommended by the World Health Organization [23], and as a nonpharmacological intervention, it is considered to be an alternative treatment for dementia [10][11][12][13][14]. In a recent animal study, improvement in cognition and hippocampal synaptic plasticity induced by acupuncture was achieved via activation of dopamine D1/D5 receptors [24].
After adjusting for geographic characteristics, region of residence and selected comorbidities, the proportion of men with diagnosis of stroke was a bit higher than that of women in both cohorts (53.48% vs. 46.52%) during this 3-year follow-up. The selected comorbidities were more prevalent in the stroke cohorts compared with the non-stroke cohorts. The mean age of the patients in the cohorts was 66.02 ± 11.72 years. In another study, older adults, residents in southern and eastern regions, patients with low insurance range, and antiplatelet use were prominent risk factors of post-stroke dementia in Taiwan [25].
We demonstrated that patients with stroke had signi cantly higher incidence of dementia rates within 3year follow-up, and the incidence of dementia in male patients with stroke was higher than that of female patients with stroke. A pooled analysis of international data classi ed important contributors and risk factors to post-stroke dementia, and age over 65 and female gender are the mentioned risk factors [26,27]. The mean age in our study and the result that stroke patients had higher incidence of dementia match with this analysis, however the effect of gender did not. Although there was no statistical signi cance, the proportion of male patients with diagnosis of stroke in our study was higher. Li CH et.al claimed that patients with stroke were at signi cantly more risk of dementia during the 5-and 10-year follow-up periods in Taiwan [25]. Post-stroke dementia is often recognized in the rst weeks to months post-ictus, and thereafter, prevalence of post-stroke dementia increases with time [28]. Ethnic differences and a longer follow-up period should be considered.
The relationship between stroke and neurodegenerative diseases, such as AD and VD, has been established [1,[2][3][4][5]. VD is the second most common cause of cognitive decline, with only AD being more prevalent [29]. Multicellular interactions within the neurovascular unit, including damage to the bloodbrain barrier, neuronal cell death or degeneration, glial reaction, and immune cell in ltration contribute to the aetiology [30]. Our results indicated stroke patients with subsequent acupuncture treatment were less likely to develop several subtypes of dementia during the 3-year follow-up. Notably, the adjusted HR for unspeci ed dementia in patients with acupuncture was 0.49 (95% CI, 0.41-0.59; p < 0.001). Pathological mechanisms of AD are poorly understood, and approximately 70% of the risk is believed to be inherited, with many genes being involved [31]. VD can also result from other conditions besides stroke that damage blood vessels and reduce circulation. This may explain why these most common aetiologies of dementia did not show statistical signi cance in our study, although there was a trend toward a rmative e cacy.
There is no obvious evidence to support any speci c mechanism that contributes to the effect of acupuncture in decreasing the incidence of post-stroke dementia. Researchers have discovered that this therapy may reduce oxidative stress, attenuate neuronal apoptosis, relieve neuroin ammation, regulate glucose metabolism, modulate neurotransmitters, and improve synaptic plasticity and blood vessel function [14]. Interestingly, we discovered evidence that revealed that those patients who received acupuncture were at a lower risk of developing dementia regardless of whether they had the diagnosis of stroke or not. Patients who choose acupuncture as a complementary treatment may place more emphasis on personal health maintenance with regard to better knowledge and attitudes for disease prevention. These factors could also contribute to reduced incidence of dementia.
A retrospective, nationwide cohort was employed in our study, which provided a large sample size and more evidence compared to case-control or cross-sectional study designs. Problems like insu cient power and the effect of selection biases were minimized. However, our study had several limitations. First, the insurance claims data we collected lacked information regarding clinical risk scores (such as the National Institutes of Health Stroke Scale or the Barthel Index), characteristics of the lesion, its location, size, laboratory examination results, and factors of patients' lifestyle such as cigarette smoking, alcohol consumption, and physical activity. Second, this study could not validate the actual acupoints used in the treatment of the stroke patients, as limited information is provided by the National Health Insurance Research Database. The acupoints chosen by traditional Chinese medicine physicians for stroke patients vary, and the bene cial effects are somewhat different [32]. Further studies need to be carried out to identify the protective effects against dementia from individual acupoints. Finally, residual confounding data is always present, even though we had utilized multivariate regression adjustments for several potential confounding factors in our analysis.

Conclusions
From the results of this nationwide retrospective cohort study, stroke patients with subsequent acupuncture treatment have a signi cantly lower risk of post-stroke dementia after adjustments for selective confounding factors. This study also showed that in general, patients treated with acupuncture had a lower risk of newly diagnosed dementia. The bene cial effect of acupuncture was investigated in our study, yet still very little is known regarding the mechanisms of acupuncture. Care must be taken in extrapolating from our results, and further investigations that overcome the limitations of this study are essential to ascertain the mechanisms for such effects and provide more evidence of the association. We analyzed administrative claims data obtained from Taiwan's NHIRD. The details of database generation, monitoring, and maintenance are published online by the Taiwan's National Health Research Institutes (http://nhird.nhri.org.tw/). Because the National Health Insurance (NHI) program is the sole payer in Taiwan, the data represents the original medical claims for all islanders covered by the NHI program, and are distributed by sex, age, or amount of average payroll-related insurance payments. The subjects of this study were recruited by reviewing monthly patient discharge data released by NHIRD, and data in the NHIRD that could be used to identify patients is scrambled before being sent to the National Health Research Institutes(NHRI) for database construction. Because the analysis was limited to aggregate secondary data that could not be used to identify the patients, the study was approved by the NHRI Ethics Review Committee, Taiwan. However, the study protocol conformed to the ethical standards established by the Declaration of Helsinki (1964), which do not require written or verbal consent for datalinkage studies.