ACCORD MIND 2014.

Study characteristics
Methods	Multisite study. Randomised, double 2x2 factorial trial. All participants entered into intensive versus standard hypoglycaemic treatment trial plus either BP or lipid trial. Randomisation via coordinating centre using unique, computer‐generated sequences using permuted blocks of 4, 8 or 12 participants. Participants in BP trial randomised to open‐label, intensive versus standard BP lowering treatment. Prespecified subgroup considered cognitive outcomes. Study duration: 40 months.
Participants	Geographical area: North America. Of the 6 American and 1 Canadian clinical centre networks in the overall ACCORD study, 6, including 54 clinics, participated in ACCORD MIND. ACCORD study criteria (n = 10,251): Type II diabetic patients with HbA1c ≥ 7.5%, aged 40 ‐ 79 years, at high risk for cardiovascular disease events. BP study criteria (n = 4,733): SBP 130‐180 mmHg with minimal proteinuria. ACCORD MIND study criteria (n = 2,977): age ≥ 55. All participants recruited into ACCORD MIND within 45 days of randomisation into ACCORD study. ACCORD MIND participants recruited between August 2003 and December 2005. Participants randomised to BP or lipid arm. N = 1,439 participants were entered into the ACCORD MIND BP trial, n = 745 randomised to the intensive BP lowering arm, n = 694 to the standard BP lowering arm.
Interventions	Visit schedule differed between the intensive and standard BP measurement groups. Participants in the intensive BP group were seen at least monthly until SBP < 120 mmHg achieved at, minimum, months 1, 2, 3 and 4, and two‐monthly thereafter. Milepost visits scheduled at 4‐month intervals for first 2 years with a different class of antihypertensive added if systolic BP ≥ 120 mmHg. Participants in the standard BP lowering group had therapy up‐titrated if SBP was ≥ 160 mmHg on one visit or ≥ 140mmHg on two successive visits. Therapy was down‐titrated if SBP was < 130 mmHg on one visit or < 135mmHg on two successive visits.
Outcomes	Analysed in review: digit symbol substitution test (DSST); Mini‐Mental State Examination; Rey Auditory Verbal Learning Test; Stroop test; Health Related Quality of Life as measured using the Medical Outcomes Study 36‐item short form health survey, Diabetes Treatment Satisfaction Questionnaire and Patient Health Questionnaire‐9. Not analysed in review: total deaths; deaths due to CVD; non fatal myocardial infarction (MI), non fatal stroke; revascularisation; heart failure hospitalisations; microvascular composite outcomes; biochemical analyses; Patient Health Questionnaire; self‐reported ability to manage diabetes therapy questionnaire.
Notes	Exclusion criteria overall ACCORD study included: recent significant hypoglycaemic event; BMI > 45; serum creatinine > 1.5 mg/dL; active liver disease; cardiovascular event or procedure within previous 3 months; weight loss; malignancy; 'any condition that, in the judgement of clinical study staff members, would preclude full participation in the study e.g. preexisting clinical evidence of dementia, substance abuse'. Of the n = 745 randomised to the intensive BP lowering arm and n = 694 to the standard BP lowering arm, n = 618 (of n = 659) and n = 649 (of n = 690), respectively, had 40 month data reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: 'unique randomisation sequences for ACCORD were computer generated centrally at the coordinating centre using permuted blocks of 4, 8 or 12 participants' 'baseline characteristics were compared between intervention groups...characteristics that differed between groups were adjusted for in post‐hoc analyses'
Allocation concealment (selection bias)	Low risk	Quote: ' unique randomisation sequences for ACCORD were computer generated centrally at the coordinating centre using permuted blocks of 4, 8 or 12 participants'
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: '(the lipid trial) is the only masked intervention in ACCORD'.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: '(the lipid trial) is the only masked intervention in ACCORD'. Quality control measures are described for cognitive testing but no mention made of blinding the assessors.
Incomplete outcome data (attrition bias) All outcomes	High risk	Imputation methods for missing data reported. Numbers deceased, lost to follow‐up and with missing assessments reported (12% missing or deceased intensive group, 11% standard group). Of the 745 and 694 in the intensive and standard BP intervention groups, 690/740 ad 659/690 with baseline DSST measurements were included in the final analyses. Those missing follow‐up data were significantly older, had higher SBP and lower DSST scores at baseline.
Selective reporting (reporting bias)	Low risk	40‐month BP measurements reported for 86% of the randomised participants (83% intensive group, 88% standard group). 40‐month results reported for DSST, the prespecified outcome, but for a selection only of the secondary cognitive outcome measures, stating that 'mean 40‐month cognitive function did not differ between intervention groups for any of the other 3 cognitive tests'.
Other bias	Unclear risk	Only included participants with type 2 diabetes mellitus. Quote: 'All participants receive nutrition and physical activity counselling, as well as a recommendation to use aspirin daily...current smokers receive smoking cessation counselling...information on current guidelines for lipids and blood pressure treatment is provided to participants' personal physicians' Quote: 'regular visit schedules differ by treatment group assignment' schedule for intensive group described more thoroughly than that for standard group, so magnitude of difference unclear. Visit effect accounted for in statistical analyses. Participants had already been randomised to intensive or standard BP management when they were invited to participate in the MIND substudy. To account for this, differences in baseline characteristics between the two groups were analysed and corrected for in post‐hoc analyses. Add‐on antihypertensive therapy was required for one participant each in the ARB and diuretic groups and three in the ACE inhibitor group.