Across 136,781 US adults with confirmed type 2 diabetes in linked health records, remission (two HbA1c readings below 6.5% after stopping all glucose-lowering drugs) was documented in 4.33% of previously treated adults and 15.57% of those untreated at entry. Median first remission lasted 226 days in treated adults. Within five years, remission ended with treatment restart in 47.83%, glycemic relapse in 31.94%, and death in 4.65%. Heavier drug burden at baseline predicted lower remission odds, falling from 4.65% with one drug class to 2.02% with four or more.
These figures sit well below the dramatic remission rates seen in intensive trials such as DiRECT, where structured weight loss and bariatric surgery studies produced remission in a large share of participants. Routine care looks very different: remission is rare, and the finding that restarted prescriptions outnumber HbA1c relapse suggests clinicians often resume medication before glucose rises, perhaps out of caution or for cardiovascular protection. That means HbA1c-only relapse definitions likely understate how fragile remission is. Limitations matter: EHR data cannot show whether patients lost weight, and treatment may be recorded imperfectly across systems. Observational records also cannot separate true remission from missed measurements or care elsewhere. This is a preprint that has not been peer reviewed, so numbers may shift. The practical message for adults with diabetes is that remission is attainable but needs sustained effort, ideally medically supervised. Incremental but valuable real-world calibration of trial optimism.