The Prognostic Value of Traditional Chinese Medicine Symptoms in Acute Ischemic Stroke: A Pilot Study

Background Stroke scales of traditional Chinese medicine (SSTCM) are promoted for use in the early prognosis. The current lines of evidence to support their performance evaluation are uneven. This pilot study aimed to investigate the correlation between traditional Chinese medicine (TCM) symptoms in the early stages of acute ischemic stroke and the prognosis of motor dysfunction through one-year of follow-up. Methods Three hundred and fifteen patients were retrospected at Longhua Hospital from January 2016 to December 2017. All patients had received standard treatments combined with acupuncture therapy, including both electroacupuncture and scalp acupuncture for a median course of five months. The observed outcomes were the Fugl-Meyer assessment (FMA), the modified Barthel index (MBI), and the modified Rankin scale (mRS) at one-year follow-up after stroke onset by multiple linear regression analysis combined with ROC curves. Results The favorable outcome rate was 74.3%, with the recurrence rate of 20.3% in the follow-up. In multiple linear regression, 10 TCM symptoms (MBI regression model) were related to the prognosis of MBI (DW 1.409, Ad. R2 0.654) and 10 TCM symptoms (FMA regression model) were related to the FMA outcome (DW 1.446, Ad R2 0.620). The two models were selected to have nine repeated symptoms (repeated model). In the ROC curves, the three models were compared with the NIHSS score, and the MBI regression model reflected the highest efficiency. Conclusions The combination of 10 TCM symptoms, once onset occurred, including hemiplegia, restlessness, hemianesthesia, short breath, headache, constipation, night sweat, tinnitus, thirsty, and gurgling with sputum, may affect the recovery of motor dysfunction. Furthermore, the improvements of TCM symptoms dynamically after treatment would be observed in a large prospective cohort. This trial is registered with NCT01806233.


Background
Stroke is associated with the highest disability-adjusted lifeyears (DALYs) cost, as the first fatal disease in China, with over two million new cases annually [1], accounting for 40% of the global total [2]. With the change of environment and unhealthy lifestyle, the incidence among young and middleaged adults (20-64 years old) displays an increasing trend [3]. e estimated onset age of stroke in China is about 10 years earlier than that in developed countries [1]. Typically, motor dysfunction is the most severe neurological deficit after stroke, with a disability rate of as high as 60% [4]. Given that China had the highest morbidity and mortality rates globally, more vigorous and effective actions are needed to reduce the burden of stroke.
Considering the high disability and recurrence rates of ischemic stroke, early prognosis plays an important role in making the treatment plans and evaluating clinical efficacy. With the progress of bio-psycho-social model of medicine, health-related quality of life (hr-QOL) and patient-reported outcomes (PROs) are regarded as the gold standards for evaluating the quality of life and subjective feelings of patients [5]. SSTCM, established under the guidance of TCM, and through the combined use of the four diagnostic methods, pay more attention to the unity of the body and spirit, pattern differentiation, and personal physique so as to adapt to Chinese conditions. However, the current evidence showed the performance evaluation of TCM scales is uneven [6]. Scales in general accepted standards have not been applied as early predictive indicators of stroke outcome in the prognostic scale [7]. Typically, some early prognostic symptoms are important for the development of precision medicine, quality of life, and prognosis of stroke, which need to be testified in the progress of medical practice.
With the rapid development of big data, real-world evidence is playing an increasing role in healthcare decisions. As a complementary role of clinical evidence, realworld data can come from various clinical practices directly, without deliberate selection of subjects and excessive manual intervention [8]. Consequently, real-world studies can reconstruct the real clinical scene maximumly in order to investigate clinical problems, especially in etiologic and prognoses research. is retrospective study, as one part of real-word studies, can provide evidence for evidence-based practice.
is pilot study aimed to investigate the correlation between TCM symptoms in the early stages of acute ischemic stroke and the prognosis of motor dysfunction. e clinical cases were retrospectively reviewed to assess the motor dysfunction in FMA through a one-year follow-up so as to provide evidence for the establishment of prognostic SSTCM.

Study Design.
In this retrospective study, the inpatient electronic medical records database at Longhua Hospital Affiliated to Shanghai University of Traditional Chinese Medicine was accessed to analyze the secondary data.

Study
Subjects. 1650 cases with acute ischemic stroke were recruited between January 1 st , 2016, and December 31 st , 2017, from five inpatient departments in Longhua Hospital: department of TCM, neurology department, emergency ward, special ward, and geriatrics department. All patients received the standard treatments in accordance with the Chinese guidelines for the diagnosis and treatment of acute ischemic stroke [9].

Diagnostic Criteria
(1) Ischemic stroke was diagnosed with reference to the criteria of cerebral arterial thrombosis from the Chinese Stroke Society [9] (2) TCM diagnostic criteria of apoplexy referred to the "Standards of Syndrome-Differentiated Diagnosis of Apoplexy" in Chinese medicine [10]

Inclusion Criteria
(1) Adults patients should meet the diagnostic criteria of acute ischemic stroke and apoplexy mentioned previously (2) Ischemic stroke can be confirmed by a brain computed tomography (CT) or magnetic resonance imaging (MRI) scan (3) Patients who experienced an acute stroke within 24 h were included.

Intervention.
In this retrospective cohort study, all patients received standard treatments [11], including basic treatments for vital signs, antiplatelet medication, intravenous thrombolytics, and mechanical thrombectomy if eligible. Additionally, acupuncture therapy and rehabilitation were also carried out at the acute phase. Acupuncture therapies, including electroacupuncture and scalp acupuncture, were also given at the acupoints of Hegu-LI4, Shenting-DU24, Fenglong-ST40, and Baihui-Du20 as well as the motor area of the scalp for 30 min once a day (5 days/week) in the hospital. Specifically, the electroacupuncture at the paretic side was divided into two protocols, including Jianyu-LI15, Quchi-LI11, Waiguan-TE5, Yanglingquan-GB34, and Zusanli-ST36 with dilatational wave (1.5∼2.5 Hz) protocol in the presence of flaccid paralysis; Tianjing-TE10, Naohui-TE13, Waiguan-TE5, Weizhong-BL40, and Chengshan-BL57 with dilatational wave (2 Hz) protocol for spastic paralysis. Sterile disposable needles (0.25 mm × 40 mm in size) were utilized in this study, which were inserted into the scalp acupoints as described in our prior protocol [12]. Briefly, the needles were twirled for 1 min (at least 200 turns per min) for once every 10 min.

Outcome Measurement.
e primary outcome was the FMA scale at one-year follow-up after stroke onset. e FMA scale is a preferred, reliable, and stable tool for the assessment of motor functional performance of patients in restorative stroke trials [13], including flexor synergy, extensor synergy, movement combining synergy, movement out of synergy, coordination, and speed. e total store of 100 reflects a normal motor function. e secondary outcomes were the MBI scale and the mRS scale at one-year follow-up after stroke onset. e MBI is a scale that measures ten basic activities of daily life, including dressing, feeding, continence of the bladder and bowel, going to the toilet, grooming, bathing, moving, walking at 45 m, and walking up and down stairs, which are related to self-care and mobility [14]. e normal scale is 100. In addition, the mRS scale is commonly used to measure the disability degree of patients with stroke or other neurological disability in daily activities, which has become the most extensively utilized clinical outcome measure in clinical studies [15]. e scale score ranges from 0-6 points. Favorable outcome was defined as a score from 0 to 2 on mRS at first year. e quantitative data such as FMA, MBI, and NIHSS were expressed as means with standard deviation.
e count data such as TCM symptoms were expressed as counts (percentage). Frequency comparisons were made with the methods of Chi-square test. Multiple regression analysis was used to determine the correlation between TCM symptoms and FMA. All variables were force-entered. Collinearity diagnostics were made to analyze multicollinearity. Durbin-Watson (DW) test was made to check the regression models. Moreover, a receiver operating characteristic (ROC) curve was drawn to verify the discriminatory power of the models.

General Patient Characteristics.
Eventually, 315 out of the 1650 patients at five inpatient departments of Longhua Hospital Affiliated to Shanghai university of TCM were enrolled in this retrospective cohort study in accordance with the inclusion and exclusion criteria from January 2016 to December 2017 ( Figure 1).
In medication therapy, 25 (7.9%) patients received thrombolytic therapy. In antiplatelet medications, 33.8% took bayaspirin, 58.6% took clopidogrel, and 17.9% took ticagrelor. 10.3% especially took two kinds simultaneously. After discharge, the patients continued acupuncture treatment in the outpatient department. e median course of acupuncture treatment was five (five to six) months.
In the follow-up, the mean scores of FMA and MBI at the first year were 83.25 ± 14.73 and 81.79 ± 13.63. Favorable outcome (n � 234, 74.3%) was defined as a score from zero to two on mRS at the first year. No patients died. e overall recurrence rate was 20.3%.

e Correlation between TCM Symptoms and Motor Function at One-Year after Stroke.
Twenty-five TCM symptoms on admission were screened totally. Using Chisquare test, 18 TCM symptoms were related with mRS at the first year, which showed their sensitivity to predict motor function of mRS (P < 0.05) ( Table 1).
In multivariable linear regression analysis, 10 symptoms were proved to be statistically correlated with MBI at the first year (P < 0.05). e DW statistic was 1.409, which indicated nonautocorrelation. e R-squared value was 0.682, which indicated the goodness of fit. e adjusted R-squared was 0.654, which showed the regression model was reliable. Figure 2 showed the analysis of residual. Figure 3 showed a linear trend between the measured and predicted values of regression. e results of the standardized B value suggested that seven symptoms might be the primary symptoms to affect the prognosis of motor function, such as restlessness, hemiplegia, short breath, headache, and hemianesthesia ( Table 2).
On the other hand, 10 symptoms were proved to be statistically correlated with FMA at the first year (P < 0.05) in multivariable linear regression analysis. e DW statistic was 1.446, which indicated nonautocorrelation. e Rsquared value was 0.650, which indicated the quality of linear fit. e adjusted R-squared was 0.620, which showed the regression model was reliable. Figure 4 showed the analysis of residual. Figure 5 indicated a linear trend between the FMA and predicted values of regression. e results of the standardized B value suggested that seven symptoms might be the primary symptoms to affect the prognosis of motor function, such as restlessness, hemiplegia, short breath, headache, and constipation (Table 3).
After screening the above regression models (MBI and FMA), nine symptoms were of repeated occurrence, including hemiplegia, hemianesthesia, night sweats, short breath, headache, tinnitus, thirsty, constipation, and restlessness, which composed the third regression model. Moreover, the total scores of these three models were calculated, in contrast to the NIHSS scores at admission,   Figure 6 showed all the four curves were in the upper left corner closely. e results of area under the curve (AUC) showed that the sequence from high to low was MBI, FMA, repeated, and NIHSS, which indicated the higher accuracy of the three TCM scales. In sensitivity, the MBI model was the best (63%). In specificity, the FMA model was the best. Conclusively, the MBI model with 10 symptoms had the highest value in the prognosis of motor function (Table 4).

Discussion
In this paper, multivariable linear regression analysis was used to determine the correlation between 25 obtained TCM symptoms at admission and motor function by means of MBI and FMA at the first year. Moreover, ROC curves were drawn to verify the efficiency of the regression models. e NIHSS scale is widely used to assess neurological impairment and evaluate the prognosis of motor function in  Evidence-Based Complementary and Alternative Medicine acute stroke as well. e NIHSS scale consists of 11 items, including consciousness, eye movement, visual field, facial symmetry, body movement, mutual movement, sensation, and language function, with good validity and reliability [16]. However, some important factors closely related to prognosis are neglected, which might reduce the efficiency in evaluation [17]. e results indicated three TCM models with reasonable accuracy to have ROC curves in the upper left triangle above the reference line ( Figure 6). Generally, the AUC can be thought of an indicator of overall "accuracy." Considering an AUC of 0.767, the model of MBI with 10 symptoms had the highest value and best discrimination, although the specificity of the model of FMA is higher. Since the primary intended use of the prognostic test is either "ruling-in" or "ruling-out" a target condition, the balance between false-positive and false-negative rates is clinically important. In order to find the optimal trade-off between sensitivity and specificity, a value of ≥8 represents this point  Generally, the ability to predict outcome in patients with acute stroke is of research and clinical value [18]. Previous stroke, age, and urinary or bowel incontinence on admission are adverse prognostic indicators of function [19]. Particularly, the severity of paresis, presence of hemianopia, reduced leg power, and size of brain lesion are all predictive of walking within 30 days after stroke [18]. Moreover, urinary incontinence, severity of hemiplegia, comorbidity, consciousness at admission, cognitive status, and depression are independent factors that are associated with the outcome in terms of activities of daily living (ADL) beyond six months [20]. e establishment of an adequate prognosis will increase the efficiency of stroke services and reduce costs. TCM symptom, as a kind of PROs, has been widely used in clinical practice for many years. e initial severity of specific symptoms and accompanied symptoms observed in the first day after stroke are important indicators of outcome at one year after stroke in this study.
Our previous studies [21,22] found that wind syndrome and phlegm syndrome are the major single syndromes occurring within 30 days of onset. is study made further investigation into the prognostic value of definite symptoms occurring on the first day. For instance, gurgling with sputum, which is a reflection of phlegm syndrome, might be a factor affecting the recovery of motor dysfunction. A previous study [7] used the Pearson correlation test and logistic regression model to review 489 patients with acute ischemic stroke and finally screened eight items of the scale, indicating that early appearance of anxiety might predict a bad outcome at the ninetieth (90 th ) day (OR: 3.17). Our study observed the recovery of motor function at the first year, which was much longer. e symptoms collected were  Some limitations should be noted in this study. Considering that the study was retrospective, the quantity of the collected information depended on the quality of medical records. Inadequate information made it difficult to assess principal component analysis for classification. However, these TCM symptoms at the early stage of stroke might be valuable for the establishment of prognostic scales. Besides, we could not observe the improvements of TCM symptoms dynamically after treatment due to the limitation of the literature. e regression model here just reflected the first day of admission. We will focus on this part in further research.

Conclusion
It is observed in this study that ,through the one-year followup, once the ischemic stroke onset occurred, the combination of 10 TCM symptoms, including hemiplegia, restlessness, hemianesthesia, short breath, headache, constipation, night sweat, tinnitus, thirsty and gurgling with sputum, may affect the recovery of motor dysfunction. Of them, seven symptoms, including restlessness, hemiplegia, short breath, headache, hemianesthesia, constipation, and gurgling with sputum, might be the primary symptoms.

ADL:
Activities of daily living AUC: Area under the curve DALYs: Disability-adjusted life-years DW: Durbin-watson FMA: Fugl-Meyer assessment GCS: Glasgow coma scale hr-QOL: Health-related quality of life MBI: Modified Barthel index MMSE: Mini-mental state examination mRS: Modified Rankin scale NIHSS: National Institutes of Health Stroke Scale PROs: Patient-reported outcomes ROC: Receiver operating characteristic SPSS: Statistical product and service solutions SSTCM: Stroke scales of traditional Chinese medicine TCM: Traditional Chinese medicine.

Data Availability
All data included in this study are available upon request through contacting the corresponding author.
Ethical Approval e study was approved by the Ethics Committee of Longhua Hospital Affiliated to Shanghai University of TCM (2014KL-006).

Conflicts of Interest
e authors declare that they have no competing interests.