OBJECTIVES: To explore the utility of a novel Subgroup Identification Tool (SIT) in a  healthcare database; specifically to identify which patient subgroup would  achieve better outcome from which treatment option.     METHODS: For the purposes of this study, two cohorts of patients with type II diabetes were extracted from the UK General Practice Research Database.  Study patients were ≥40 years old, were newly prescribed with antidiabetics Drug A or Drug B, and had at least 6-month pre-index and 12-month post-index history. The index date was defined as the date of first prescription for Drug A or Drug B.  The outcome was the average HbA1c from 3 months post-index to the end of 12-month follow-up.  Subgroups were constructed using the SIT, which employs a novel SIDES (subgroup identification based on differential effect search) methodology. RESULTS: A total of 4824 patients were identified initiating Drug A and 1007 patients initiating Drug B. Slightly more patients achieved HbA1c ≤7% for Drug A (46.4%) compared with Drug B (42.6%), translating to the number needed to treat (NNT) of 26 in favor of Drug A.  The SIT identified a subgroup (male, ≤ 71 years old without a prescription for antihypertensives) where the Drug A patients responded more favorably than Drug B in terms of achieving HbA1c ≤7% (NNT= 8 in favor of Drug A); and a subgroup (female, low density cholesterol <120mg/dl with a prescription for angiotensin-converting enzyme inhibitors) where the Drug-B patients responded more favorably (NNT= 10 in favor of Drug B). CONCLUSIONS: This study indicates that the SIT can be useful when applied to healthcare data to identify subgroups that are more likely to achieve better outcome from one of the two comparative treatment options. The results from the SIT could be used for hypothesis generation.  Validation using independent data is warranted.