BACKGROUND: With a large percentage of population-based epidemiological research still using paper-based methods as the primary data collection tool, there is potential for improving efficiency through web-based data collection. The Tennessee Department of Health (TDH) utilized the Research Electronic Data Capture (REDCap) software to design a web-based case-control study of community-associated Clostridium difficile infection. The application, developed at Vanderbilt University, was used to construct a database modeled after computer-assisted telephone interviewing in which the computer presents the questions to the interviewer in the appropriately programmed sequence. The database eliminated the need for double data entry and tracked call metrics for each interviewer.
METHODS: The case-control study included 10 instruments containing over 700 data fields. The database was created to not only manage interview data, but to serve as a mechanism to track interviewer productivity. A call log tracking form was created which captured: date/time of call, time of day, number of attempts, call duration, and outcome (e.g., completed, refused, ineligible, unreachable, invalid number). Data quality rules were implemented within REDCap to ensure that incorrect values/responses were flagged immediately during the interview. This virtually eliminated the need for data cleaning. Reports were developed within REDCap to track the number of calls in progress, time of day calls were made, outcome of each attempt, and number of cases/controls enrolled. De-identified data were exported out of REDCap and submitted to the Centers for Disease Control and Prevention through secure upload.
RESULTS: From October 2014 to March 2015, 983 people were contacted (87 possible cases and 896 possible controls). Five interviewers made 2,658 call attempts (287 to cases and 2,371 to controls) to enroll 39 cases and 26 controls. An average of 7.4 calls was required to recruit one case and 91.2 calls to recruit one control. Each individual was called on average 2.7 times before a final outcome was achieved. Over 50% of the controls could not be enrolled due to invalid telephone numbers from the commercial database of landline numbers obtained, or no one in the household was in the appropriate sex/age group. Approximately 90% of the cases without a matched control were under 40 years old (p < .001).
CONCLUSIONS: Utilizing REDCap to implement real-time data collection and call tracking for a case control study has allowed TDH to: quantify the effort involved in recruitment, evaluate the quality of the commercial database used for control selection, and optimize productivity for future studies.