Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

Health Services Research & Development

Veterans Crisis Line Badge
Go to the ORD website
Go to the QUERI website
2015 Conference Logo



2015 HSR&D/QUERI National Conference Abstract

Printable View

3150 — Research Application of Adaptive Double Data Entry Procedures

Hostetter TA, Rocky Mountain MIRECC; Forster JE, Rocky Mountain MIRECC; Schneider AL, Rocky Mountain MIRECC; Huggins J, Rocky Mountain MIRECC; Brenner LA, Rocky Mountain MIRECC;

Objectives:
Kleinman (2001) developed an approach for quality checking data entry, called Adaptive Double Data EntRy (ADDER), as an alternative to traditional double data entry. ADDER is designed to enhance the efficiency of data checking while maximizing the number of errors caught, thus it is essential that implementation procedures also be efficient. This approach utilizes a separate probability-based algorithm for each data entry person to determine which data forms will be reentered, and is a function of the number of errors detected in previous forms. Our objective was to develop these procedures and implement ADDER at the Rocky Mountain MIRECC (Mental Illness Research, Education and Clinical Center).

Methods:
The first step to implement ADDER is to ensure the Access database is set up to automatically capture all required ADDER fields prior to the start of data entry. After all data are entered into the database, queries are created using SQL code to sort the forms and to assign each an ADDER order number. To determine a baseline error-rate, the first 15 forms identified in the ADDER table are re-entered into a second database and compared in SAS using a macro that reports the errors detected. The number of errors are verified and entered into an R function that performs the ADDER algorithm and reports the next set of 15 forms to be re-entered. This iterative process continues until all ADDER forms have had the opportunity to go through the algorithm. A number of parameters within the algorithm are adjustable such that ADDER can be customized to each study's needs.

Results:
Within our center, ADDER has been utilized for 5 studies to date. On average, 22.8% of study forms were re-entered. The minimum was 17.2%, detecting 31 total errors on 335 forms and the maximum was 29.4%, detecting 65 total errors on 182 forms.

Implications:
Although there is a time commitment for initial procedure development, the resulting process provides increased confidence in the data checking process.

Impacts:
Utilizing ADDER allows researchers to be assured that they have completed enough double data entry of their research data to be confident in the quality of their data.