We are writing to inform the readership of a recently discovered error in our data that alters the results of 2 articles recently published in the ARCHIVES.1,2 The error involves a single variable—the duration of the longest episode (recorded in days or weeks) of major depression (MD) as assessed at personal interview. In the original data verification program, this variable had 3 characters, allowing responses of up to 999 days or weeks. At this stage, one of us (C.O.G.) examined this and the other variables used in our diagnostic algorithm for major depression and noted nothing unusual. Subsequently, we interviewed several twins who reported MD with a duration of more than 999 days or weeks. The data verification program was changed to add a fourth character to the field but, critically, this information was not conveyed to Dr Gardner. In the data set prepared for analysis, for data entered after the change in format, the rightmost character was stripped from this variable. For example, a response of 90 days became 9 days. Therefore, any twin who met the other DSM-III-R criteria for MD3 but recorded a maximal duration of illness between 2 and 20 weeks (or 14 and 140 days) was incorrectly classified as unaffected. We failed to detect this error in our data analyses.
Kendler KS, Gardner CO, Prescott CA. Corrections to 2 Prior Published Articles. Arch Gen Psychiatry. 2000;57(1):95–97. doi:10.1001/archpsyc.57.1.94
Artificial Intelligence Resource Center
Customize your JAMA Network experience by selecting one or more topics from the list below.