| 1 BackgroundAccording to the type of data source, meta-analysis methods could be divided to aggregate patient data (APD) and individual patient data (IPD) meta-analysis.APD meta-analysis refers to combine and analysis quantitative evidence from related published studies to produce results. Traditional meta-analysis always used aggregate study level data obtained from study publications or study authors. IPD meta-analysis aims to summarize the evidence on a particular clinical question from data recorded for each participant.Since 1980’s, in the medical and health field, systematic review and meta analysis have become the cornerstone of evidence-based medicine and one of the best evidence, and APD meta-analysis was widely used in evaluation of curative effect of Traditional Chinese Medicine. But Traditional Chinese Medicine has the characteristics of individual syndrome differentiation and complex intervention, which either was handled as heterogeneity of the data and not included in the meta-analysis or merged into "Traditional Chinese Medicine treatment" or "acupuncture treatment". The results of meta-analysis can not only reflect the characteristics of Traditioanal Chinese Medicine, but also the conclusion is uncertain, which could not provide guidance for clinical practice.IPD meta-analysis is considered to be the "gold standard" of systematic review. In 1991, the first IPD meta-analysis published for evaluation of platinum based chemotherapy in the treatment of advanced ovarian cancer effect in BMJ in the word. Since 2000, IPD meta-analyses were published increasingly, which mainly focused on studies in cancer, cardiovascular disease, diabetes, hepatitis, pulmonary inflammation and other fields. At present, in Traditional Chinese Medicine and complementary and alternative medicine research field, only 3 IPD meta-analysis articles were published. In China, there was no published article which applied IPD meta-analysis.There is a big difference between Traditional Chinese Medicine and Western medicine in the evaluation of clinical efficacy. Whether IPD meta-analysis is suitable for evaluating the efficacy of Traditional Chinese Medicine, what are the advantages and difficulties in IPD meta-analysis when comparing with the APD meta-analysis, and which questions are worth for studying, would be the key points.2 ObjectiveTo analyze research methods of IPD meta-analysis in intervention studies, explore the characteristics of IPD meta-analysis of Traditional Chinese Medicine and the advantages, and difficulties of IPD meta-analysis compared with APD meta-analysis and provide methodological reference for future research to select and use meta-analysis method in the field of Traditional Chinese Medicine.3 Method3.1 Literature reviewWe did a comprehensive search on published IPD meta-analysis/systematic reviews for intervention studies in four journals for 2010-2014 including "LANCET", "BMJ", "JAMA", "Cochrane". We investigated the research field, the method on the quality evaluation method and statistical analysis method of IPD meta-analysis.3.2 Case studyIPD meta-analysis was adopted to evaluate the effect of Traditional Chinese Medicine treatment of common hand foot mouth disease (HFMD), including the following steps:1) We did systematic electronic research and hand search, initially identified qualified published studies, then connected published study author to get the original patient data.2) We assessed the methodological quality of published trials according to 5 domains based on the Cochrane handbook and checked every original patient data from authors.3) We combined the original data of each trial into a unified database; 4) We redefined uniform analysis population, standard outcome definition, then calculated the outcome according to the above definition for every trials, finally we polled each results into a analysis.3.3 comparative studyFirst, In order to ensure the comparability, we re-analyzed the published literature into IPD meta-analysis using APD meta-analysis. Second, based on the results, we compared of results of two methods of meta-analysis.4 Results4.1 The results of Literature researchA total of 40 articles of IPD meta-analysis were included,38 of which is systematic reviews, and two were meta-analysis.16 (40%) articles were organized by the collaboration.Research fields:IPD meta-analysis were most widely used in cancer with 11 (37.5%) articles, followed by cardiovascular with 8 (20%) articles, Surgery Dept with 5(12.5%) articles and nervous system with 5 (12.5%) articles.Methods of quality evaluation:according to the existing literature by published literature evaluation and the original data were evaluated two methods were used to quality evaluation, including published literature bias risk assessment and the original data checking. For published literature bias risk assessment, literature published in Cochrane often used Cochrane quality assessment recommended standard. The original data checking was no uniform standard, usually focus on the balance of the problem, the implementation of random, blind method, the consistency of published literature research results and result of recalculating IPD, the balance of the key variables of the baseline.The principle and method of IPD meta-analysis:29 (72.5%) of 40 IPD meta-analysis articles reported by the principle of intention to treat principle and 1 (2.5%) articles using modified ITT principle. For evaluation of curative effect,19 (47.5%) studies used "two step" method,11 (27.5%) studies uses "one-step" method and 8 (20%) studied used two methods. There are three ways of IPD meta-analysis for survival time data analysis.33 studies focused on the covariate effect including three main types using the "one-step" method and two types using the "one-step" method.4.2 The results of TCM treatment for common hand foot and mouth disease using the IPD meta-analysisAfter electronic and hand searches, we contacted the with published literatures authors.6 RCTs trials supported by 2009 national support projects were enrolled, including 2609 individual patients.2 RCTs were included comparing the Traditional Chinese Medicine and placebo.4 RCTs compared the TCM, western medicine and TCM plus western medicine.Comparison of TCM plus symptomatic treatment and placebo plus symptomatic treatment:there are 687 individual patient data include in ITT population analysis. Meta analysis results show that two groups of rash disappearance time of herpes had statistical significance (HR:1.54,95%CI:1.25 to 1.90; I2=0.0%), Traditional Chinese Medicine group is better than the placebo group. Subgroup analysis results showed no significant difference between TCM and placebo group in different age, sex, baseline temperature and WBC levels patients.Comparison of TCM and Western medicine:there were 1274 individual patient data include in ITT population analysis. Meta analysis showed there was no statistical significance between two groups in the incidence of severe, cumulative time for body temperature recover, disease cure rate and cumulative time for rash disappear time.Comparison of TCM plus Western medicine treatment and Western medicine alone:there were 1280 individual patient data include in ITT population analysis. Meta analysis showed there was statistical significance between two groups in the incidence of severe (RR:0.36,95%CI:0.19 to 0.67; I2=0.0%) and cumulative time for body temperature recover (HR:1.35,95%CI:1.03 to 1.76; I2=64.7%). For incidence of severe, subgroup analysis results show no significant difference between two groups in different age, sex, baseline temperature and WBC levels patients. For cumulative time for body temperature recover, subgroup analysis results showed no significant difference between two group in different sex, baseline temperature and WBC levels patients, but In younger cases (age≤3), there was significant difference between two groups (HR:1.48,95%CI:1.16 to 1.90; I2=39.2%), TCM plus western medicine treatment was better than western medicine alone, and in older case (age>3) there was no statistically significant differences among two treatments. There was statistically significant (P=0.03) between different age subgroups, which indicate two kinds of treatment were different in different age subgroups.4.3 The results of comparing the APD and IPD meta-analysis4.3.1 Comparison of the number of included studies/literatureFor same 2009 national supported projects, IPD meta-analysis included 6 trials and a total of 2609 patient cases. APD meta-analysis included 8 RCT trials and 1571 patient cases.4.3.2 Comparison of quality assessment resultsTo compare of individual patient data checking and published literature reports, we found 2009 national support projects have some problems of publication bias, duplication, incomplete outcome report and selective outcome reporting etc. We also accessed the risk of bias using Cochrane risk of bias tool for eight publish trials. Each access using information from either trial reports alone or trial reports with additional information collected for IPD meta-analyses. The results found that a total of 29 (72.5%) low risk of bias,5 (12.5%) unclear, bias and 6 (15%) high risk of bias based on publications alone, comparing 20 (50%) low risk of bias,1 (2.5%) unclear, bias and 19 (47.5%) high risk of bias based on publications and addition information from IPD.4.3.3 Comparison of treatment effectsDifference of indicators:Because of the information insufficient or inconsistent from published literature, there are only the 4 outcomes were pooled in meta-analysis, but 16 outcomes were polled in IPD meta-analysis. From the comparison of the number of participants in two meta-analyses, because IPD meta-analysis recalculated and reanalyzed outcomes according to the same standard, the number of patients in IPD meta-analysis was more than APD meta-analysis.Difference of treatment effects:· For rash herpes disappear time, IPD meta analysis showed no statistical significance between the Traditional Chinese Medicine group and Western medicine group, but APD and meta analysis results show significantly shortened of Traditional Chinese Medicine group than western medicine group.· IPD meta-analysis showed that compared with single western medicine treatment, Traditional Chinese Medicine and Western medicine combined can significantly reduce the incidence of severe cases, the recovery time of temperature. But APD meta-analysis showed no statistically significant differences for treatment temperature returned to normal time between Traditional Chinese Medicine and Western medicine combined treatment and single western medicine treatment.Difference of statistical analysis method:· Due to the limitation from published literatures report, APD meta-analysis didn’t report HR and 95%CI for survival time data.· In IPD meta analysis, the subgroup meta-analysis was perform by different age, sex, baseline temperature and WBC levels, while APD meta analysis couldn’t carry out these subgroup meta-analysis based on baseline characteristics.· IPD meta-analysis also carried out a sensitivity analysis based on PPS population, which couldn’t be performing by APD meta-analysis.5 ConclusionsIPD meta-analysis is suitable for clinical research of Traditional Chinese Medicine. Compared with APD meta-analysis, IPD meta-analysis can avoid publication bias; improve data quality, the data for statistical heterogeneity homogenization process and the use of more flexible statistical analysis to the maximum extent.Although IPD meta-analysis has many advantages, APD meta-analysis will cost more manpower and time.There are characteristics of many kinds of interventions and diversified composite indicators of the effect of TCM clinical research, IPD meta analysis was applied to the clinical evaluation of Traditional Chinese Medicine can be screened out the advantages of Traditional Chinese Medicine interventions and outcomes by subgroup analysis method, but still based on the premise of more number of original clinical research accumulation and implementation of interventions and terms of TCM coding standards.In order to promote the IPD meta-analysis study, we need to accelerate the clinical study registration system, transparent clinical study results, built clinical trial data sharing platform and follow the international clinical data interchange standards (CDSIC) standards to design clinical trials. |
PDF Full Text Request