Error minimization strategy in clinical and translational studies (CTS) through the development of a data management routine (DMR)

Abstract

17567 Background: Clinical studies are the basis for the decision making process in health care. So quality assurance is mandatory. In CTS, of intervention or descriptive nature, the use of spreadsheets as databases is frequent among investigators favoring mistakes during data entry, mainly due to the matricial distribution of data, urging strategies to error minimization. Creation of complex systems by informatics technologists is viable, but expensive and unnecessary for medium and small sized studies. Otherwise, there are systems that allow databases to be built by non-programmers upon training. But the customization of those tools requires knowledge on CTS. Our aim wasto ensure the quality of CTS conducted at Instituto Nacional de Cancer of Brazil (INCA) through the development of a DMR. Methods: DMR is based upon MS Word 2003, MS Access 2003 and SPSS 13.0. Content variables were defined based upon studies running at Clinical and Translational Research Division (CTRD). For its development, 8 different studies (retrospective, molecular correlation, formal clinical trials) were used. Results: We generated a DMR based upon two pillars: a) a multi-step standard operational procedure that includes meetings with the investigators and data collection training; establishment of clinical report forms supported by guidelines for its use; a pivotal pilot data collection for identification and clarification of eventual misunderstanding; a revision of 30% of the cases enrolled; b) a backbone relational database customized for each study by health professionals trained in our service with a friendly easy-to-use interface and several validation rules to reduce data entry mistakes. It stores, organizes, encodes data, filters eligibility criteria, calculates data, generates reports to investigators for proper follow-up and generates spreadsheets easily imported by SPSS for statistical analysis. Since 2004 this routine has been applied to all the studies referred to CTRD, decreasing the intra-mural historical 15% mistake rate in patient inclusion to nearly 0%. Conclusions: In a low-cost and simple manner an effective DMR has been generated, decreasing the likelihood of mistakes with great impact on the quality of the data generated in CTS at INCA. No significant financial relationships to disclose.