Fig. 2.

Adherence measurements as assed by self-reporting and sachet tools compared with MEMS device. Panel a Data were not available for 17 children because 15 self-reporting forms were either missing or not filled out and records of 2 MEMS devices missing because 1 device was damaged and another was not returned at end of 30 days after randomization. Panel b Data were not available for 15 children because 13 children lost or did not return empty sachets and 2 MEMS devices were unavailable. Not all data points in Panel a and Panel b are visible due to over-lapping of adherence counts. The red line indicates perfect agreement in adherence counts between pairs of assessment methods. For each panel, data points presented above the reference line indicate overestimation of adherence assessed by self-report or sachet count, respectively, as compared to adherence measured by MEMS, whereas data points presented below the reference line indicate underestimation of adherence measured by MEMS