ACCORDION MIND: results of the observational extension of the ACCORD MIND randomised trial

The Memory in Diabetes (MIND) substudy of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study, a double 2x2 factorial parallel-group randomised clinical trial, tested whether intensive compared with standard management of hyperglycaemia, BP or lipid levels reduced cognitive decline and brain atrophy in 2977 people with type 2 diabetes. We describe the results of the observational extension study, ACCORDION MIND (ClinicalTrials.gov registration no. NCT00182910), which aimed to measure the long-term effects of the three ACCORD interventions on cognitive and brain structure outcomes approximately 4 years after the trial ended. Participants (mean diabetes duration 10 years; mean age 62 years at baseline) received a fourth cognitive assessment and a third brain MRI, targeted at 80 months post-randomisation. Primary outcomes were performance on the Digit Symbol Substitution Test (DSST) and total brain volume (TBV). The contrast of primary interest compared glycaemic intervention groups at the ACCORDION visit; secondary contrasts were the BP and lipid interventions. Of the surviving ACCORD participants eligible for ACCORDION MIND, 1328 (68%) were re-examined at the ACCORDION follow-up visit, approximately 47 months after the intensive glycaemia intervention was stopped. The significant differences in therapeutic targets for each of the three interventions (glycaemic, BP and lipid) were not sustained. We found no significant difference in 80 month mean change from baseline in DSST scores or in TBV between the glycaemic intervention groups, or the BP and lipid interventions. Sensitivity analyses of the sites with >= 70% participation at 80 months revealed consistent results. The ACCORD interventions did not result in long-term beneficial or adverse effects on cognitive or brain MRI outcomes at approximately 80 months follow-up. Loss of separation in therapeutic targets between treatment arms and loss to follow-up may have contributed to the lack of detectable long-term effects. ClinicalTrials.gov NCT00182910.