Fourteen studies involving 29,319 people with diabetes were included and 11 studies involving 29,141 people were included in our meta-analyses. Treatment duration was 56.7 months on average (range 6 months to 10 years). Studies included people with a range of kidney function. Incomplete reporting of key methodological details resulted in uncertain risks of bias in many studies. Using GRADE assessment, we had moderate confidence in the effects of glucose lowering strategies on ESKD, all-cause mortality, myocardial infarction, and progressive protein leakage by kidney disease and low or very low confidence in effects of treatment on death related to cardiovascular complications and doubling of serum creatinine (SCr).
For the primary outcomes, tight glycaemic control may make little or no difference to doubling of SCr compared with standard control (4 studies, 26,874 participants: RR 0.84, 95% CI 0.64 to 1.11; I2= 73%, low certainty evidence), development of ESKD (4 studies, 23,332 participants: RR 0.62, 95% CI 0.34 to 1.12; I2= 52%; low certainty evidence), all-cause mortality (9 studies, 29,094 participants: RR 0.99, 95% CI 0.86 to 1.13; I2= 50%; moderate certainty evidence), cardiovascular mortality (6 studies, 23,673 participants: RR 1.19, 95% CI 0.73 to 1.92; I2= 85%; low certainty evidence), or sudden death (4 studies, 5913 participants: RR 0.82, 95% CI 0.26 to 2.57; I2= 85%; very low certainty evidence). People who received treatment to achieve tighter glycaemic control probably experienced lower risks of non-fatal myocardial infarction (5 studies, 25,596 participants: RR 0.82, 95% CI 0.67 to 0.99; I2= 46%, moderate certainty evidence), onset of microalbuminuria (4 studies, 19,846 participants: RR 0.82, 95% CI 0.71 to 0.93; I2= 61%, moderate certainty evidence), and progression of microalbuminuria (5 studies, 13,266 participants: RR 0.59, 95% CI 0.38 to 0.93; I2= 75%, moderate certainty evidence). In absolute terms, tight versus standard glucose control treatment in 1,000 adults would lead to between zero and two people avoiding non-fatal myocardial infarction, while seven adults would avoid experiencing new-onset albuminuria and two would avoid worsening albuminuria.