We appreciate Dr. Saleem’s (1) interest and comments and thank the editor for the opportunity to clarify the points raised about our study (2).
Our study was an effectiveness trial. The target population for the Annual Diabetes Assessment Program (ADAP) included all managed care members with diabetes assigned to the participating provider groups. The enrollment criteria were pragmatic, excluding those without diabetes, those who were no longer being followed by the primary care physician, and those who were deemed “unsuitable candidates for the study” by their primary care physicians. We obtained consent from both participants and their primary care physicians to comply with good clinical practice (3). The attrition in the intervention group from 173 eligible individuals to 103 enrolled subjects reflected those who did not consent to the ADAP. The decrease in sample size from year 1 to year 2 (from 103 to 83 in the intervention group and from 111 to 71 in the comparison group) was largely due to subjects’ inability to come for their year 2 appointments (e.g., moved away, too ill, etc.). The drop-out rate for the intervention group was 20% (20 of 103), not 52%. We analyzed outcomes for study completers to avoid within-group and between-group bias in interpreting the results. The composition of the study population reflected the racial and ethnic composition of the health plan. We saw no reason to oversample Hispanics. Even if we had done so, we would not be able to generate a sample size sufficient to draw inferences about that population.
The intervention was added to usual care, and as a result, the ADAP was more expensive than usual care alone. Since the ADAP resulted in no improvement in outcomes, usual care dominated the experimental intervention (4). Three registered nurses/certified diabetes educators implemented the ADAP. Clearly, it was feasible to implement the ADAP with nonphysician providers. The question now is how to structure the ADAP to improve both processes of care and intermediate outcomes. Our study was limited in that the ADAP generated only patient and provider feedback. As combinations of interventions have been shown to be more effective in producing change (5,6), future studies should include the ADAP and additional interventions such as nurse case management and more effective tracking and reminder systems to impact both processes and outcomes of care.
