Negative or neutral treatment effects reported in many clinical trials reflect average treatment effects in the trial population and may mask heterogeneous treatment effects across patient subgroups. Machine learning (ML) methods are valuable tools to study these treatment heterogeneities. Using ML methods and data collected in the ACCORD trial, we identified patient subgroups that might benefit from intensive glycemic and blood pressure intervention. The ACCORD found that intensive glucose (A1c <6.0% vs. A1c 7.0% to 7.9%) and SBP (<120 mmHg vs. <140 mmHg) control did not significantly reduce CVD risks. We reanalyzed data from 10,251 participants from the glucose trial and 4,733 from the SBP sub-trial separately. We used Major Adverse Cardiovascular Events (MACE) as our CVD outcome. We applied causal forest (CF) and causal tree (CT) models to identify participant characteristics that modulate the effectiveness of intensive glycemic and SBP intervention from a list of 48 candidate variables (demographics, disease histories, medications, and biomarkers) at the study baseline. Compared to controls, intensive SBP intervention reduced MACE in those with normal HDL (women >55 mg/dL, men >45 mg/dL; relative risk [RR] 0.51, 95% CI: 0.34-0.74%) and those with LDL >100 mg/dL (RR: 0.74, 95% CI: 0.59-0.94) , but had no significant effects in other patient groups. Intensive glycemic control reduced MACE in those with baseline A1c <8.5% (RR: 0.79, 95% CI: 0.67-0.93) and those with glomerular filtration rate >1mL/min/1.73m2 (i.e., highest quartile; RR: 0.74, 95% CI: 0.55-0.99) ; reductions were not significant in other patient groups. Our findings may assist clinicians in developing a more precise approach to managing persons with type 2 diabetes at high risk of CVD events. Future clinical trials may be needed to confirm these findings.


H.Kianmehr: None. P.Zhang: None. N.Singh ospina: None. L.Shi: None. V.Fonseca: Consultant; Abbott, Asahi Kasei Corporation, Bayer AG, Novo Nordisk, Sanofi, Research Support; Fractyl Health, Inc., Jaguar Gene Therapy, Stock/Shareholder; Abbott, Amgen Inc., BRAVO4Health, Mellitus Health. J.Guo: None. H.Shao: Board Member; BRAVO4HEALTH, LLC.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at