| 1. ObjectiveTake insomnia as an example, to evaluate the effect of physicians’treatment for insomnia by TCM patterns differentiation, and to analyze the difference and the characteristics of TCM physicians’treatment.Based on the results of former two, the methodology was explored and established to evaluatethe effect of treatment by theTCM patterns differentiation with the application of physicians’cohort study.2.Method2.1 Study designThe prospective longitudinal cohort study for each TCM physicianswith all their samples was applied and established. The physicians’cohorts should be recommended by the professional committee of TCM on Sleep medicine, should represent their regional characteristics, and should ask the TCM physiciansto be famous for their treatment on insomnia.2.2 TCM physicians and ParticipantsThe DSM-V was usedfor diagnosis. The participants were included if they complained insomnia as chief complaint, met the diagnosis of insomnia disorders, had age from 18 to 65,and accepted the informed consent. The participants were excludedif they had extremely severe diseases,or uncontrolled diseases, orsevere depression, or psychiatric problems with suicide trends, or pregnancy and lactation.Sample size was as large as the total number of samples that the TCM physicians met and treated at their outpatients within one year. Each physician should at least collect 100 participants for analysis.Inclusion criteria of physicians were that physicians should treat insomnia with TCMmore than 20 years, and the groups of patients were fixed.Clinic is equipped with research assistants and computers that can help TCM physicians include the qualified patients. TCM Physicians had tendency to join the trial.2.3 OutcomesMain outcomes selected by experts were effective ratesof PSQI and TST (the criteria for PSQI efficacyreferred to that the total scores were≤7 points orthe total scores declined≥4points;but the criteria for PSQI inefficacyreferred to that the total scores declined< 4points. And TST efficacyreferred to thattotal time was≥7 hours or improvement rate was ≥15%;TSTinefficacyreferred to that the improvement rate was <15%).2.4 InterventionsInterventions meant theTCM physicians’individualized treatment of prescription drugs, combination treatment,decocting and taking methods, etc. the frequency and the duration of course for patients were determined by the TCM physicians according to the patient individualized health condition.Patients visited their TCM physicians at their outpatient clinics. The control or explosion of cohort included the outcomes’ changes of before and after treatment for patients, andoutcomes’comparison between TCM physicians.2.5 Data analysesData analyses included three parts:evaluating the effect of treatment based on the TCM patterns differentiation between TCM physicians, finding the characteristics of each TCM physician and their TCM patterns differentiation, comparingthe differences between individual TCM physicians and all the TCM physicians.2.5.1 Comparison of the effect of treatment based on the TCM patterns differentiation between TCM physiciansThe main information of cohort study was described, and the logistic regression analysis was applied to compare the effect of treatment based on the TCM patterns differentiation between TCM physicians. Besides, the logistic regression analysis proved that the classification of insomnia symptoms and the four cohorts were interacted and may be the main factor to explain the differences of the effect, which had been assumed by the experts that classification of insomnia symptoms may explain the characteristics of each TCM physician. Thus, the subgroup analyses of insomnia classification were applied for the further analyses of the characteristics of each TCM physician. However, due to the limited samples and the large differences among the subgroups, the comparative results were only described.Two methods were used to analyzethe factors affecting the changes of outcomes. The outcomes TST and PSQI effective rates were used as the index for the efficacy group and the inefficacy group, and then the multivariable logistic regression was carried out to select the factors with P<0.05. Another mehod compared the information at baseline between cases of the patients’loss after the first visit and those of the patients more than 2 visits, with an assumption that all the loss cases were the inefficacy cases. Thus, the index of differences between those groups maybe the factors affecting the outcomes.2.5.2 Relationship of TCM patterns, individualized treatment and changes of outcomes was analyzed by the data mining with 6 steps:Step 1, Java software was used to calculate the similarity, Jaccard coefficient, between thedata of TCM physicians’texts prescription and the data from clinical practical database. According to the Jaccard coefficient distribution, threshold was made to determine the parameters of community partition; and then Gephi 0.8.2 beta software was used for the community partition method to analyze the distribution characteristics of Jaccard coefficient; higher similarity of drug combination represented and grouped the similar prescription like TCM physicians’ones.Step 2, core herbs and add-or-subtract herbs were analyzed by mining the similar prescription. Abstract the visits’ID of the patients from the similar prescription, and then Liquorice hierarchicalnetwork analysis system (with adjusting parameters:Layer Number=3, Degree Coefficient=2) was applied for the three levels of core herb combination structures. The first-level core herb combinations structures were regarded as the most important structures or the prescriptions, other two levels were usually regarded as the add-or-subtract herbs.Step 3, the characteristics of people was analyzed by mining the similar prescription. Step of similarity analysis,31 variables were selected as the main population characteristics by the expert consensus, and then the principal component analysis (PCA) by SPSS was used to select and weight the variables, which showed great contribution to the similarity. Community partition method by Gephi 0.8.2 beta software was used to analyze the characteristics of the community or module underneath the population from the similar prescription. The number of nodes, number of edges,values of weighted edge, values of module degrees and values of average degrees determined the division of population.T-tests and chi square testswere calculated by SPSS and EXCEL, and P values were adjusted according to the number of variables.Step 4, the Liquorice hierarchicalnetwork analysis system was also used to calculate the Pointwise Mutual Information(PMI) of herbs-TCM patterns and herbs-symptoms, which represented the relationship of herbs and the population characteristics; higher values, stronger relationship. The paper only included the value of PMI≥1.00.Step 5,analyze the relationship of the herb structures and their corresponding characteristics of population. The top 2 frequency of TCM patterns and the top 20 frequency of TCM symptoms were selected as the population characteristics to the first-level herb combinations structures, regarded as the core structure of prescription. As to the rules of add-or-subtract herb, the differences of population characteristics (including TCM patterns and symptoms) between community and the whole population from the patients ID of similar prescription were first determined. If no significant differences were found, it meant there was only one population corresponding to the core structures of prescription. If two or more than two communities could be divided from the whole population, it meant that the characteristics such as TCM patterns and symptoms (but expect the characteristics represented the core structures) representing the communities were corresponding to some herbs. In such case, the relationship rules of the population characteristics and the add-or-subtract herbs (second and third level structures) could be determined by the PMI values, and all the herbs listed after the unique characteristics with high to low rank. Sometimes, due to the overlap of characteristics both corresponding to the core structure and others, there was no results for the add-or-subtract herbs.Step 6, Relationship of TCM patterns, individualized treatment and changes of outcomeswere determined by determining thephysicians’TCM patterns differentiation and treatment protocol. Abstract the patient ID from the similar prescription, and then extracted the information according to the checklist.2.5.3The comparison of the contents of TCM patterns, individualized treatment and changes of outcomes.Compare the correlation of symptoms-herbs, symptoms-TCM patterns, and symptoms-treatment principle between the patients of individualized TCM physicians and those of either two TCM physicians. The calculation method included quantitative and qualitative methods. The former one calculated the similarityJaccard coefficients of symptoms, herbs, TCM patterns, and treatment principles; and then calculated the Pearson correlation coefficients ofsymptoms-herbs, symptoms-TCM patterns, and symptoms-treatment; if P was< 0.05, the correlation was statistically significant. The qualitative correlation analysis refers to calculate the correlation of the Jaccard coefficients of symptoms and the overlap ratio of any others, such as herbs, TCM patterns, and treatment principles. The overlap ratio referredto that when it was in the interval of values of the symptoms’similarity, the number of pairs who could be calculated with Jaccard value divided by all the number of the pairs in theintervals.With such results, the relationship of TCM patterns, individualized treatment and changes of outcomes were compared between the individualized TCM physician and the results from determining, and were compared between any two TCM physicians.2.6 Protocol implementation and quality controlProtocol implementation and data collection and management includedstudy design, design, testing and optimization of structured electronic medical record templates, analysis of clinical data, data preprocessing, analyses plan, data mining, and interpretation of the results and guiding the clinical practices.Quality control included the standardized operation process (SOP) documents and ensured the implementation of the above documents.2.7 The study has been approved by the Institutional Review Board of the Institute of Basic Research in ClinicalMedicine, China Academy of Chinese Medical Sciences. The study should prevent the rights of patients as well as the included TCM physicians.The study protocol was registered at the ClinicalTrials.gov (No. GZR81230086).3.Results3.1 Summary of the cohort studyFour TCM physicians’cohorts wereset up for 1 year, whose regional distributions werein Fujian, Hubei, Henan and Shandong. Screening cases were 962 (1417 visits), 359 cases were included in the data analyses (777 visits). At baseline, the assessment of patients’ sleep showed thatthe total sleep time (TST) in the Cohort 2 was the least;the patients in the Cohort 1 had severe problems in the Waking time during sleep (WTDS) and Early morning awaking in sleep (EMAIS);and patients in Cohort 4had severe problems in sleep efficiency (SE).Among four cohorts, PSQI showedthat the quality of all the patients’sleep was poor, and the ISI and CGI-S showed most patients were mild insomnia. The high rank symptoms were the difficulty falling asleep, maintaining sleep disorders, the degree of worry about sleep, fatigue after wake up.Mental factors were the main cause of insomnia.3.2 Description about the content of TCM patterns, individualized treatment and changes of outcomesA total of 31 TCM patterns were collected by four TCM physicians, and the TCM Physician A collected 11, the TCM Physician B18, the TCM Physician C 4, and the TCM Physician D 6. A total of 93 symptoms were collected by four TCM physicians, and the TCM Physician A collected 69, the TCM Physician B 70, the TCM Physician C 56, and the TCM Physician D 65. A total of 19 TCM treatment principles were collected by four TCM physicians, and the TCM Physician A collected 12, the TCM Physician B 26, the TCM Physician C 12, and the TCM Physician D 20. A total of 32 prescriptions were applied by four TCM physicians, and the TCM Physician A collected 6, the TCM Physician B8, the TCM Physician C 5, and the TCM Physician D 13. A total of 191 herbs were applied by four TCM physicians, and the TCM Physician A collected 71, the TCM Physician B83, the TCM Physician C 71, and the TCM Physician D 86.Effective rates of all TCM physicians were from 27.50% to 100.00%, where EMAISE showed the lowest effective rate, while HAMA highest effective rate; and the TCM physician C showed higher effective rates among the entire outcome index. The major outcome index effective rates of TST and PSQI from high to low ranks were TCM Physician C, D, B, A.3.3 Comparison of effective rates among TCM physiciansA total of 14 covariables from baseline such as insomnia symptom classification, insomnia etiology classification, PSQI, TST and other ten covariables were selected into the logistic model. The results showed that significant difference happened in the TST and PSQI among all four TCM physicians when four cohorts were compared (P<0.05). Pairwise comparisons were conducted between either of TCM physician with the adjusted P value equal to 0.008. The results showed that the TCM Physician C had significantly better effect than any one (P<0.001), but there was no significant differences between either comparisons of other physicians.Because the classification of insomnia symptoms significantly contributed to the outcomes, the results of subgroup analyses showed the tendency that the subgroup of difficulty falling asleep showed better effect thanthe subgroup of maintaining sleep disorders, and the subgroup of mixed sleep disorders of difficulty falling asleep and maintaining sleep showed the worst effect. However, the classification of insomnia causes did not show the contribution to the effect.3.4 Analyses of the factors influencing the effective ratesThe comparison of the cases with patients’loss and those received treatment more than 2 visits showed that the significant differences happened in the patients’ demography information and the insomnia-related symptoms, but less in the TCM pattern related symptoms. The comparison of the population with inefficacy and efficacy of TST and PSQI groups showed that the negative correlation of insomnia related symptoms or cause of insomnia with the effect. That meant the insomnia related symptoms were the main factors influencing the effective rates. However, the TCM patterns related symptoms were complex to the effective rates, like the negative correlation of thin, diarrhea and bad appetite with effective rates in the treatment of TCM Physician A and B. The TCM pattern of deficiency of heart and spleen, snoring and irritability were positivelycorrelated with the TCM Physician C, but negative for upsets. The Maihua, Maiku, irritability and bitter were positively correlated with TCM Physician D, but the negative for Maiqing.3.5 Relationship of TCM patterns, individualized treatment and changes of outcomesResults of data mining took TCM Physician A as an example that three core prescriptions were analyzed based on those that the TCM physician offered. TCM patterns, individualized treatment and changes of outcomes were obtained. Besides, the qualitative comparisons were conduct between the results of data mining and those the TCMphysician offered. All above results could prove that the methodology of cohort study design, complement, data mining and analyses was scientific and feasible.3.6 The rules of TCM patterns, individualized treatment and changes of outcomes amongthe patients of individualized TCM physicians and those of either two TCM physiciansThe correlation of TCM patterns, individualized treatment and changes of outcomes from TCM Physician A was stronger than those from either two TCM physicians, which proved that the TCM Physician A showed high self-consistence in the relationship of TCM patterns, individualized treatment and changes of outcomes; but there were large differences between either two TCM physicians. All the results proved the research hypothesis that TCM patterns differentiation took TCM physicians as the core part, and was an individualized diagnosis and treatment process of high correlation of TCM patterns, individualized treatment and changes of outcomes.3.7 Basic methods for evaluating the TCM patterns differentiation and treatmentBased on the analyses of research process and the key questions, and comparison of assessing the intervention as the core and the physician as the core, the basic methods for evaluating the TCM patterns differentiation and treatment were explored. 4. ConclusionTCM physicians as the core of the cohort study method can evaluate the effect of different physicians’treatment based on their TCM patterns differentiation.Based on the effectiveness of the TCM physicians’treatment, the regularity and characteristics of their treatment by TCM patterns differentiation were analyzed.That method was scientific and feasible to the research hypothesis that TCM patterns differentiation took TCM physicians as the core part, and was an individualized diagnosis and treatment process of high correlation of TCM patterns, individualized treatment and changes of outcomes.The innovation included the theoretical innovation and the methodological innovation. Thetheoretical innovation was to put forward TCM physicians as the core of the TCM patterns differentiation and treatment evaluation method. Methodological innovation was to provide a protocol, design and implementation of the core of cohort study with the physiciansas the core; to provide a structured electronic medical records system; and to provide data analyses methods of TCM patters-treatment-outcome relationship; and to provide the framework of clinical key elements for the TCM patterns differentiation and treatment.Some problems were raised in the research that each cohort had many cases dropped once they completed the first visit, which limited the analyses of the whole sample. Due to the limited time, long time follow-up was not conducted in the research, and with the same reasons, research only included the insomnia patients with some mental problems, which limit the subgroup analyses of the characteristics of TCM physicians’treatment. The result need to be interpreted and expressed in aobvious and easy-catch methods.Perspectives are toapply the knowledge map to interpreter the study results.The individualized clinical path of each TCM physician was made by the results of the study. RCT was used for assessing the efficacy and safety of one prescription that was selected from the study results, showing determining relationship of TCM patterns, individualized treatment and outcomes. The checklist of the TCM patterns differentiation and treatment need moreextensive expert consensus. |
PDF Full Text Request