Factors associated with compliance and non-compliance by physicians in a large-scale randomized clinical trial

In order to minimize the amount of incomplete follow-up data, reducing the non-compliance of participating physicians is one of the key issues for the data coordinating center in a multi-center trial. Identifying the physicians' non-compliance in advance is considered to be an important strategy for more efficient conduct of trials. In this study, we identified physicians' characteristics and factors associated with the need for individual visits to institutions to collect data or to complete information during two years of follow-up in a large Japanese investigator-initiated trial related to cardiovascular disease.

We categorized the physicians into two groups, "complier" and "non-complier". Odds ratios and corresponding 95% confidence intervals were calculated for 11 factors related to the characteristics of and compliance by physicians. Multiple logistic regression analysis was also performed. In addition, we evaluated the incremental cost for obtaining additional information of the non-compliant physicians.

Three factors were identified in multiple logistic regression analysis as being significantly associated with compliance status: 1) prior participation in clinical trials (OR = 0.40 95%CI = 0.21-0.74); 2) physician opinion that the support system for case registration and follow-up was well organized (OR = 0.41 95%CI = 0.22-0.75); and 3) number of patients recruited (OR = 2.25 95%CI = 1.01-5.02). The actual incremental cost was about US $112,000 (14.4% of total routine follow-up costs) for the non-compliant physicians during the 2 years, or about US $570 per patient.

Investigator-initiated clinical trials have recently attracted great interest, but they often suffer from insufficient funding. If trial networks are to be well organized, it is important that trials are conducted more efficiently. We believe that our findings will be useful for reducing the additional burden associated with incomplete follow-up data and data lost to follow-up when planning future trials.