Almudena Areosa Sastre, Robin WM Vernooij, Magali González‐Colaço Harmand, Gabriel Martínez

Prevention of cognitive impairment and dementia is an important public health goal. Epidemiological evidence shows a relationship between cognitive impairment and Type 2 diabetes mellitus. The risk of dementia increases with duration of disease. This updated systematic review investigated the effect on cognitive function of the type of treatment and level of metabolic control in people with Type 2 diabetes. To assess the effects of different strategies for managing Type 2 diabetes mellitus on cognitive function and the incidence of dementia. We searched ALOIS (the Specialized Register of the Cochrane Dementia and Cognitive Improvement Group (CDCIG)), the Cochrane Library, MEDLINE, Embase, PsycINFO, CINAHL and LILACS on 15 October 2016. ALOIS contains records from all major health care databases, (CENTRAL, MEDLINE, Embase, PsycINFO, CINAHL, LILACS), as well as from many trials' registers and grey literature sources. We included randomised controlled trials (RCTs) which compared two or more different treatments for Type 2 diabetes mellitus and in which cognitive function was measured at baseline and after treatment. Two review authors independently extracted data and assessed the quality of the included RCTs. We pooled data for comparable trials and estimated the effects of treatment by using risk ratios (RRs) and mean differences (MDs), according to the nature of the outcome. We assessed the quality of the evidence using GRADE methods. We identified seven eligible studies but only four provided data we could include in efficacy analyses. Two of these studies compared intensive versus standard glycaemic control and two compared different pharmacological treatments. All studies were at unclear risk of bias in at least two domains and one large study was at high risk of performance and detection bias.(a) Two studies with 13,934 participants at high cardiovascular risk provided efficacy data on intensive versus standard glycaemic control. A third study with 1791 participants provided additional data on hypoglycaemic episodes and mortality. There is probably no difference between treatment groups in the number of participants who decline by at least 3 points on the Mini-Mental State Examination (MMSE) over five years (RR 0.98, 95% CI 0.88 to 1.08; 1 study; n = 11,140; moderate-quality evidence); and there may also be little or no difference in the incidence of dementia (RR 1.27, 95% CI 0.87 to 1.85; 1 study; n = 11,140; low-quality evidence). From another study, there was probably little or no difference in MMSE score after 40 months (MD -0.01, 95% CI -0.18 to 0.16; 1 study; n = 2794; moderate quality evidence). Participants exposed to the intensive glycaemic control strategy probably experience more episodes of severe hypoglycaemia than those who have standard treatment (RR 2.18, 95% CI 1.52 to 3.14; 2 studies; n = 12,827; moderate-quality evidence). The evidence from these trials suggests that the intensity of glycaemic control may have little or no effect on all-cause mortality (RR 0.99, 95% CI 0.87 to 1.13; 3 studies; n = 15,888; low-quality evidence).(b) One study with 156 participants compared glibenclamide (glyburide) with repaglinide. There may be a small advantage of glibenclamide on global cognitive function measured with the MMSE after 12 months (MD -0.90, 95% CI -1.68 to -0.12; low-quality evidence). No data were reported on the incidence of dementia, hypoglycaemic events or all-cause mortality.(c) One study with 145 participants compared rosiglitazone plus metformin to glibenclamide (glyburide) plus metformin over 24 weeks. It reported only on cognitive subdomains and not on global cognitive function, incidence of MCI or dementia, hypoglycaemic events or all causes of mortality. We found no good evidence that any specific treatment or treatment strategy for Type 2 diabetes can prevent or delay cognitive impairment. The best available evidence related to the comparison of intensive with standard glycaemic control strategies. Here there was moderate-quality evidence that the strategies do not differ in their effect on global cognitive functioning over 40 to 60 months.