Table 1.

Summary of REMOVAL study outcomes

Outcomes	Baseline (mean ± SD)	Difference or ratio (metformin vs placebo)	Main effect (p value)	Treatment-by-visit interactiona (p value)	Effect of metformin over 3 years: clinical interpretation
Primary outcome
Mean cIMT(mm)	0.782 ± 0.162	−0.005	0.166	−	No significant reduction in progression of arteriosclerosis
Secondary outcomes
HbA1c (%)	8.1 ± 0.8	−0.13	0.006	0.016	Reduction at 3 months by 0.24% (2.6 mmol/mol) (p < 0.0001) but not sustained thereafter
HbA1c (mmol/mol)	64.0 ± 9.0	−1.4	0.006	0.016
LDL-cholesterol (mmol/l)	2.20 ± 0.71	−0.13	0.012	0.310	Sustained reduction by 0.13 mmol/l
eGFR (ml min−1 [1.73 m]−2)	92.0 ± 21.2	+4.0	<0.0001	0.662	Increased 4 ml min−1 (1.73 m)−2 at 3 months, then declined in parallel with placebo; requires further investigation
Retinopathy (two-step change from baseline)	−	0.76 (OR)	0.568	−	No effect
Weight (kg)	84.0 ± 14.7	−1.17	<0.0001	0.274	Sustained reduction by 1.17 kg
Insulin dose (U/kg)	0.65 ± 0.28	−0.005	0.545	0.002	After 6 months, reduced by 2 U/day (p = 0.045; post hoc analysis)
Endothelial function (reactive hyperaemia index; AU)	2.26 ± 0.74	−0.06	0.302	0.566	No significant change
Tertiary outcomes
Severe hypoglycaemia (per patient year)	0.16	1.23 (IRR)	0.442	−	No significant change
Treatment satisfaction	31.77 ± 3.94	−0.12	0.668	0.629	No significant change
Maximal common cIMT(mm)	0.918 ± 0.196	−0.013	0.0093	−	Significant reduction in progression of atherosclerosis
Occurrence of vitamin B12 deficiency (<150 pmol/l)	−	2.76 (HR)	0.0094	−	Risk of vitamin B12 deficiency more than doubled vs placebo

Data were analysed by ANCOVA other than for carotid outcomes (repeated measures regression), retinopathy (logistic regression), hypoglycaemia (negative binomial regression) and vitamin B12 (Cox proportional hazards)

All analyses were pre-specified unless otherwise stated

^aPresented for data analysed by ANCOVA only

AU, arbitrary units; eGFR, estimated GFR; IRR, incidence rate ratio