Assessment of Intermingled Phlegm and Blood Stasis Syndrome in Coronary Heart Disease: Development of a Diagnostic Scale

Background Intermingled Phlegm and Blood Stasis Syndrome (IPBSS) is a common feature in patients with coronary heart disease (CHD). In clinical practice, the diagnostic agreement of clinical doctor of Chinese Medicine (CM) is poor. We previously developed a IPBSS diagnostic scale for use by general practitioner. Objectives To assess a IPBSS diagnostic scale that we previously developed for use by non-experts. Methods This is a multicenter, prospective study involving eight study sites across China. Eligible patients were adults (≥18 years) with CHD as demonstrated by a history of myocardial infarction, stenosis, or past coronary revascularization. IPBSS was assessed using a scale that consisted of 14 items in two domains (e.g., phlegm and blood stasis). The score range for each item was 0 to 3 points. Maximum total score was 72 points. Diagnostic accuracy was verified using consensus opinion by two independent experts as reference. Results A total of 1,142 CHD patients were included. IPBSS was established in 729 subjects using the IPBSS diagnostic scale. In ROC curve analyses, at the optimal cut-off of 25.5, the sensitivity and specificity of the IPBSS scale were 67.6% and 72.4%, respectively. The area under the ROC curve was 0.741 (95%CI: 0.711-0.772). Conclusions The newly developed IPBSS scoring system showed moderate performance in diagnosing IPBSS in CHD patients. Data from further large-scale diagnostic test accuracy studies are warranted. This trial is registered with ChiCTR-OOC-15006599.


Introduction
Coronary heart disease (CHD) is one of the major causes of mortality and morbidity in developed countries [1]. Although CHD-related mortality has gradually declined over the last decades, it remains a leading cause of death in Europe [2]. In China, the number of people diagnosed with CHD has reached a total of 11 million in 2017 [3]. Many CHD patients have been successfully managed by CM, which emphasizes pattern identification [4]. Such an approach helps to identify the underlying pathological bases of the disease and is critical in selection of optimal treatment [5]. Effective integration of pattern identification from CM with differential diagnosis from Western medicine provides a solid basis to advancing the development of a model for individualized medicine [6].
The core value of pattern identification is CM syndrome, which is identified by experts through comprehensive analysis using four diagnostic methods that are unique to the practice of CM [7]. CM syndrome may evolve during disease progression. As a result, lack of simple and standardized assessment protocol represent a major obstacle in using pattern identification in daily medical practice. Efforts have been devoted to resolve this issue, including the use of hyperspectral medical tongue images for tongue diagnosis [8], tactile sensor system to detect pulse signals for pulse diagnosis [9], and scales and questionnaires developed to standardize disease diagnosis [10]. Scales and questionnaires could be divided into four types: "different disease with same syndrome" diagnostic scale of CM syndrome [11], "disease combined syndrome (multiple)" diagnostic scale of CM syndrome [12], Syndrome Factors Diagnostic scale [13], and "disease combined syndrome (single)" diagnostic scale of CM syndrome [14]. However, these diagnostic scales are typically developed without rigorous scientific grounds in item selection and often lack validation [15]. We have previously developed a diagnostic scale to assess IPBSS in CHD [16]. In the current study, we examined the performance of this scale in a group of CHD patients. (1) history of myocardial infarction (ST-elevation myocardial infarction (STEMI) or non-ST-elevation myocardial infarction (NSTEMI)); (2) >50% stenosis in at least one main branch of coronary arteries, as established by coronary angiography or computed tomography angiography; (3) history of percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) [18,19].
Subjects with unstable angina were excluded.
. . IPBSS Scale. The diagnostic scale for IPBSS was developed based on systematic literature review [20] of best available evidence via an expert consensus-based Delphi [21] and analytic hierarchy process (AHP) [22] approach. The scoring system included a total of 14 items [16] that could be classified domains into two domains (e.g., phlegm and blood stasis). Possible range of score was 0 to 3 points for each item and 0 to 72 points for the overall assessment. After assessment using the IPBSS diagnostic scale, symptom differentiation was conducted independently by two CM experts with the following qualifications: (i) specialization in cardiovascular diseases; (ii) specialization in traditional Chinese medicine or integrative Chinese and Western medicine; (iii) having a senior professional title; (iv) engagement in daily clinical practice for at least 20 years.
Diagnosis of IPBSS was confirmed only on the basis where both experts reached agreement on a particular diagnostic case. If the two experts disagreed on patient status, IPBSS was considered not to be present. The performance of the IPBSS scoring system was investigated based on the consensus of opinion from the two experts. The two experts were not involved in assessing IPBSS using the scale.
. . Statistical Analysis. Statistical analysis was performed using the SPSS statistical software (IBM Corp. Released 2015. IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY: IBM Corp.). Using the experts' diagnoses as the standard, we determined the receiver operating characteristic (ROC) curves for the IPBSS scores, the maximum Youden Index value, and the corresponding cut-off points [23]. In addition, the sensitivity, specificity, positive, and negative predictive values (PPV and NPV, respectively) were calculated using the following formulas: Sample size was calculated as described previously [16].  Table 1).

. . Patient Baseline
. . Performance of the IPBSS Scale. The total score of IPBSS scale did not conform with normal distribution (median score: 21, range 12-33). Using the ROC analysis, the optimal cut-off score of the IPBSS diagnostic scale was determined to be 25.5 (Table 2). At this cut-off, the sensitivity, specificity, PPV, and NPV of the diagnostic scale were 67.6%, 72.4%, 81.2%, and 55.9%, respectively. The area under the curve (AUC) was 0.741 (95% confidence interval (CI), 0.711-0.772) (Figure 2).  Figure 1: A flow diagram of the participants.

Discussion
The present study showed that, using a cut-off score of 25.5, our novel IPBSS scoring system produced reasonable sensitivity and specificity in differentiating patients with IPBSS from those without, using expert diagnoses as the reference. Currently, only a few CM diagnostic scales are available [24], and some of these scales do not provide clear explanation or description of the methodology of their development processes [12], and some without adequate quality control in scale design procedure, data entry [25]. Therefore, utilization of these diagnostic scales poses substantial issues in terms of consistency and accuracy.
Consequently, we conducted the current study to illustrate an effective validation method of a diagnostic scale by systematically reviewing the current literature together with a AHP-Delphi consensus method to establish the list 4 Evidence-Based Complementary and Alternative Medicine of items to be included in the IPBSS diagnostic tool, eventually reaching a total of 14 items. The items we examined were extracted from modern standards and guidelines [26,27], which were the result of wisdom by traditional Chinese medicine researchers. And the content validity of the scale depends on the quality of these standards and guidelines. Weighting of each item was achieved via the AHP. A four-tiered classification was designed to score each item (0 to 3 points), thereby providing a total score ranging from 0 to 72 points. We subsequently developed ROC curves and observed that a cut-off score of 25.5 was associated with optimal sensitivity, specificity, PPV, and NPV.
The current study has several limitations. First, we only assessed China-based participants and did not include non-CHD subjects [28]. Our study population is, therefore, not representative and whether our IPBSS scoring system could be used to assess subjects with other conditions of other ethnicities requires further investigation. Second, we prespecified opinion from two experts as the gold standard. A true "gold standard" is hard to attain in studies exploring diagnostic test accuracy of traditional Chinese medicine. Wang and Zhou proposed the use of random effects models to estimate the diagnostic performance without a gold standard by accounting for the correlation structure among different tests or practitioners [29]. However, the models proposed are in need of further investigation to fully validate their application. We opted to use the expert consensus method not only because this method is already widely used but also because it is highly unlikely for two experts to provide entirely different opinions in CM diagnoses, given that the experts we enrolled to participate were all strong candidates with solid theoretical knowledge and practical experience. Third, our diagnostic scale is yet to be verified using a large-scale, external study cohort. For future development, we anticipate to further modify the diagnostic scale with potentially less for ease of use.
In conclusion, our novel IPBSS scoring system produced moderate performance in identifying the presence of IPBSS in Chinese CHD patients. And we will improve the scoring system in future.

IPBSS:
Intermingled Phlegm and Blood Stasis Syndrome CHD: Coronary heart disease CM: Chinese medicine AHP: Analytic hierarchy process ACC: American College of Cardiology AHA: American Heart Association STEMI: ST-elevation myocardial infarction NSTEMI: Non-ST-elevation myocardial infarction PCI: Percutaneous coronary interventional CABG: Coronary artery bypass grafting CTA: Computed tomography angiography ROC: Receiver operator characteristic T2DM: Type 2 diabetes mellitus COPD: Chronic obstructive pulmonary disease AUC: Area under the curve.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Ethical Approval
The current study was approved by the Medical Ethics Committee of Tianjin University of Traditional Chinese Medicine (no. TJUTCM-EC2015000).