Background: Currently, there is no agreement among researchers on the definition of craving and its underlying theoretical model. The Obsessive-Compulsive Drinking Scale (OCDS) seems to measure certain aspects of craving, but its theoretical basis remains unclear. The aim of this study was to investigate the structure of alcohol craving, using OCDS data.
Methods: OCDS data from four studies were pooled to obtain a large and heterogeneous sample of 505 participants. All participants were treatment-seeking alcoholics meeting DSM-IV criteria for alcohol dependence. The factor structures of the OCDS previously found were evaluated using confirmatory factor analyses. The goodness of fit of these solutions was compared with those of alternative causal models: an obsessive-compulsive disorder model, an inhibition model, and a cognitive-behavioral model. These alternative models were based on modern theories about craving and were tested in the OCDS data, using structural equation modeling. In this way, the current study replaced simple correlational analysis by a more sophisticated causal way of analyzing the underlying structure of the OCDS items. The best fitting model was selected by comparing the mean discrepancy between the implied and observed matrices of the models.
Results: The data showed that the previously reported factor structures had to be rejected. Also, the inhibition model and obsessive-compulsive disorder model did not fit the data. The cognitive-behavioral model showed encouraging fit. Optimizing strategies were applied to further improve the fit of this model, which resulted in a model with close fit to the data.
Conclusions: The causal cognitive-behavioral model proved to be superior. It showed that the OCDS contains many items that do not represent the core concept of craving but instead are indicators for the consequences of craving. From this model, it seems that craving, in a narrow sense, can be reliably assessed with only five items of the OCDS.