Daw E W, Province M A, Gagnon J, Despres J P, Bouchard C, Leon A S, Skinner J S, Wilmore J H, Rao D C
Division of Biostatistics, Washington University Medical School, St. Louis, MO 63110-1093, USA.
Ann Epidemiol. 1997 Oct;7(7):452-62. doi: 10.1016/s1047-2797(97)00082-3.
The primary goal of the HERITAGE Family Study was to document the role of the genotype in the response to aerobic exercise training. Toward this end, nuclear families were enrolled in a 20-week exercise training program, with a large variety of tests performed before and after the training. Since study drift has the potential to adversely affect the results, reproducibility and potential bias over six consecutive 4-month periods were examined for selected test.
Intraclass correlations (ICC), technical errors (TE), coefficients of variation within subject (CV), and means were calculated with use of the pretraining test results for each of the six time periods. To check for homogeneity, hypothesis tests were performed on the intraclass correlations and means. If homogeneity was not found across all six periods, further tests were performed to assess differences between pairs of time periods.
There was little evidence for real drifts in reproducibility, with most tests having ICCs of 0.8 or better. Only a few tests showed any change over time, and in no case was there evidence of a systematic drift in mean values.
Overall, the reproducibility of the HERITAGE Family Study tests and assays considered in this paper was found to be very good, with no evidence of any systematic drift over time.
