Dear Dr. Faraone,
We would like to kindly respond to the letter submitted by Tomoyuki Kawada entitled “Actigraphic evaluation for patients with attention deficit/hyperactivity disorder.” This letter raised some issues claimed by the author regarding the use of the automatic device actiwatch in our study “Refining psychiatric phenotypes for response to treatment: contribution of LPHN3 in ADHD” published in your journal.
Actiwatch device was used in our work to monitor the children's motor activity, as a potentially quantitative phenotype pertinent for response to methylphenidate in ADHD children. As the author mentioned, we did not use the standard procedure (sleep/wake judgment algorithm) available in the software named Actiware®. This sleep wake algorithm was not used simply because we did not include any data on sleep in our study. However, the relevant measure for our specific research interest was the original activity count recorded.
The issue raised by the author is that activity from different actiwatch does not always agree with each other. Since different machines have been used on different children, the author claims that a calibration bias was present in our data. Although this is true that inter-machine variation can have an impact on the absolute value of activity, this effect has been implicitly taken into account when we computed the treatment effect measure for each subject, which was used as our main outcome in our analysis. The treatment effect was measured by subtracting the activity count obtained during the placebo week from the activity count from the active treatment week. Since the same actiwatch was used in both weeks for each child, the machine calibration effect is removed from our main outcome. As a result, we believe that the validation study proposed by the author is not relevant for our study.
We hope that this answer meets your expectation,
Aurelie Labbe, PhD,
Assistant professor, McGill University,
Department of Epidemiology and Biostatistics