Semi-individualised Chinese Medicine Treatment as an Adjuvant Management for Diabetic Nephropathy

This will be an open-label randomised parallel pilot clinical trial.

Sample size justification The sample size is calculated based on the target control of inflation factor. In order to be 95% confident (two-sided) that the subsequent main study actually achieves a power of 80% with nominal power set at 90% (i.e., a 10% power forfeit), the inflation factor should be less than 1.15.

At IF = 1.15, a sample size of 80 is therefore needed to have 95% one-sided confidence that the main study will achieve at least the nominal power to test the hypothesis that an add-on of the whole Chinese medicine treatment plan could be more effective in stabilising the glomerular filtration rate among diabetic nephropathy patients when compared to having just routine care.

For subgroup analysis, a sample size of 25 patients per each subgroup could achieve 80% one-sided confidence that the effect of stabilising glomerular filtration rate is different within subgroups of similar CM clinical pattern.

With 5 subgroups, a sample size of 148 patients is needed in this pilot trial to allow a 15% attrition rate.

Data management A trial management committee (TMC) formed by Principal Investigator, Co-Investigator and Research Assistant will centralise all the data of the trial. Chinese Medicine Practitioner and Research Assistant will collect, clean and send the data of patients to TMC on a daily basis. Questionnaires on clinical presentation and Chinese Medicine diagnosis will be sent, collected and cleaned by TMC directly daily by Research Assistant. If there is no response from the patients after 30 days of the last contact, Research Assistant will follow up and call the patients. All data will be double entered to computer and cleaned before analysis to prevent data entry errors. All transfer of data will include encryption and follow the guidelines by European Directive on Good Clinical Practice and adhere to Data Protection Act to protect the patients' confidentiality. TMC will have regular meetings monthly with experts to discuss the progress of the trial.

Data analysis Missing values, if any, will be imputed with regression. Patient without a postrandomisation assessment for a particular efficacy endpoint will be excluded from the analysis of that endpoint.

Regression analyses will be used to compare the adjusted mean of estimated GFR, UACR, HbA1c, FBG, FGF-23, MCP-1 and Cystatin C at week 48 between 1) the combination of all intervention groups and combination of all control groups, 2) individual treatment subgroup and its matching control group , and 3) different control groups with the corresponding baseline values as covariates. Change score analysis will be supplemented.

To minimise Type I error inflation, the analysis will follow a hierarchical approach in the order of 1) individual treatment subgroup versus its control subgroup, 2) combination of all intervention groups versus the combination of all control groups as to avoid Type I error inflation. The primary outcomes are the change of GFR and UACR.

Subgroup analysis will be performed for CKD stage 2 and 3 separately. Sensitivity analyses will be performed for 1) missing data imputed with regression, 2) missing data imputed with last-observation-carried-forward (LOCF) and 3) per protocol drop out of patient.

The adverse events will be analysed in a narrative manner. The percentage of all adverse events and the rate of attrition due to adverse events will be compared between intervention groups and control groups.