In Reply: Dr Pitrou raises 2 concerns regarding the COPE trial. First, she suggests we used a per-protocol rather than ITT analysis, which may have biased our outcomes because dropouts were not included. However, we did not use a per-protocol approach: we did not exclude caregivers who stayed in the study but refused treatment, as a per-protocol approach entails.
Because our only data before 4 months (the primary end point) were from baseline, our choices for managing missing data from dropouts were limited. We rejected last observation carried forward because of its biasing effect when both groups naturally change from baseline to follow-up. We did deviate from ITT in that families who missed follow-up assessments were not included. However, we retained the ITT principle; all study participants with follow-up data were included in analyses regardless of participation level. This approach is used in other large-scale multisite nonpharmacologic interventions with caregivers and dementia patients.1,2
Gitlin LN, Hauck WW, Winter L. Well-being of Patients With Dementia and Their Caregivers After a Biobehavioral Home-Based Intervention. JAMA. 2010;304(22):2482-2484. doi:10.1001/jama.2010.1790