OBJECTIVES: To quantify the influence of methodological assumptions/parameters on the results of prescription-based NA analysis based on 3-years prescription data (German Statutory Health Insurance Fund) covering 241,537 T2DM patients. METHODS: With the help of MPRs in an interval-based approach, a NA-base scenario for 25 anti-diabetic active ingredients was calculated for each patient. In a scenario analysis, the quantitative influence of all in all 9 parameters on the MPR level was derived. The most important parameter concerned the definition of an ideal prescription profile (“100%-adherer”). Assumptions simplifying the real prescription behaviour did not allow to accurately reflect the variety of medications and the clinical need to change medications. Therefore, a total of eight clinically meaningful prescription profiles were derived assigning patients exclusively by the use of self-developed algorithms. For each patient a MPR estimated by standard methodology (base case) was compared with the MPR based on our novel approach. RESULTS: In the base case, the average MPR resulting from the analysed active ingredient combinations was 80.76%. A total of 62.85% of patients had an MPR<80%. According to the novel prescription profiles, patients were distributed as follows: 59.9% Mono-medication, 12.9% Single-Drug Switcher, 11.3% Single-Drug Add-on, 2.0% Multiple-Drug Add-on, 5.0% Polytherapy consistent, 2.5% Polytherapy Add-on, 3.5% Polytherapy Drop-off, and 1.3% Polytherapy Switcher. A total of 1.6% of the patients could not unequivocally be assigned to one of the categories. Comparing a base-case MPR analysis with our novel approach resulted in MPR deviations in specific patient groups of up to 27.4 percentage points. CONCLUSIONS: Probably the biggest challenge in NA analysis based on prescription data is to differentiate between physician-induced and patient-induced medication changes. The first should be reflected in the adequate profile of an NA analysis and should not be misinterpreted as NA itself. The methodology described presents a powerful alternative for defining clinically meaningful prescription patterns.