Abstract
RATIONALE: One of the current models for drug (opioid) user typology employs differential access conditions to categorize takers into short-access and long-access groups (ShA, LgA) with the rationale that these groups represent distinct opioid user phenotypes. However, with the idea that differential vulnerabilities to opioid effects are already present prior to experimenter assignment into these groups, it is unclear that these groups represent distinct opioid user types. To clarify this, we have developed a method that includes principal component analysis-gaussian mixtures model clustering of variables derived from a new CENTERED (Cumulative Experience-Normalized Time-Effect on Response as an Exponential Decay structure) model. The goal of this study was to utilize CENTERED clustering to test the hypothesis that ShA and LgA groups defined by the experimenter via random assignment are composed of mixtures of individuals that belong to distinct opioid user types. METHODS: We reanalyzed data from a previous study in which the experimenter assigned male Sprague Dawley rats (n = 30) self-administering 0.1 mg/kg/infusion oxycodone for 20 days into ShA (3h-access) and LgA (9h-access). We conducted CENTERED clustering on all takers, irrespective of assigned group(s) to determine if the experimenter-assigned groups included mixtures of individuals from groups (opioid user types) identified via CENTERED clustering. RESULTS: CENTERED clustering revealed that ShA and LgA groups consisted of mixtures of different opioid user types. CONCLUSIONS: CENTERED clustering revealed that experimenter-imposed grouping via differential access conditions limits our ability to identify distinct opioid user types that already exist naturally in the population.