A web-based system for patient registering and matching in a prospective and observational clinical study

According to the principle of clinical trial design, treatment and control group should be comparable. In an ongoing prospective observational study to evaluate the effect of CRT (Cardiac resynchronization therapy) for heart failure patients, it is desired that two groups of patients (CRT treatment group and Non-CRT treatment group) were matched into pairs when being registered, and only matched patients were followed up so as to reduce the overall costs. We developed a registering and matching system to meet this requirement. Heart Failure Patients' data were collected from 44 centers by an internet-based system prospectively. 11 potential confounding variables, ranked from 2 to 3 levels, were included in this study. 5 variables were matched absolutely, while 6 others were matched by an “unbalance score”. Patients from each group respectively with the minimum difference were matched within a pre-defined threshold. The matching was done automatically by database system at every night. Unmatched patients who had been registered over 4 weeks were no longer eligible for matching. After 28 months, 478 patients were registered and 206 were matched. After matching, the balance of the covariates between two groups was improved. Alternatives of matching methods are also discussed.