Relapse as a model of nonadherence to dietary treatment of diabetes.

The generality of Marlatt and Gordon's (1980, 1985) model of relapse to dietary treatment of diabetes was tested. Forty-six adults with insulin-dependent diabetes mellitus (IDDM) and 43 obese adults with non-insulin-dependent diabetes mellitus were interviewed regarding their most recent dietary violations, and the results were coded using the schema developed by Marlatt and Gordon. As the model would predict, most nonadherence episodes occurred in a limited range of high-risk situations. Although the two diagnostic groups lapsed in remarkably similar situations, there was a tendency for the IDDM adults to report a larger proportion of lapses in situations characterized by negative emotions. Approximately 27% of the dietary lapses occurred when the person was busy with a competing activity or had no choice, and these lapses did not fit into Marlatt and Gordon's coding schema. These violations consistently differed from those studied by Marlatt and Gordon in that they were errors of omission rather than errors of commission. Evidence for the abstinence violation effect was not found in this sample. Overall, the results suggest that most nonadherence to dietary treatment of diabetes may be best understood as intermittent lapses that typically do not develop into full-blown relapses.