Bi-directional interaction between hypoglycaemia and cognitive impairment in elderly patients treated with glucose-lowering agents: a systematic review and meta-analysis

Aims: To examine the bi-directional relationship, whereby hypoglycaemia is a risk factor for dementia, and where dementia increases risk of hypoglycaemia in older patients with diabetes mellitus treated with glucose-lowering agents.  Methods: We searched MEDLINE and EMBASE over a 10-year span from 2005 to 2015 (with automated PubMed updates to August 2015) for observational studies of the association between hypoglycaemia and cognitive impairment or dementia in participants aged >55 years. Assessment of study validity was based on ascertainment of hypoglycaemia, dementia and risk of confounding. We conducted random effects inverse variance meta-analyses, and assessed heterogeneity using the I2 statistic.  Results: We screened 1177 citations, and selected 12 studies, of which nine were suitable for meta-analysis. There were a total of 1 439 818 participants, with a mean age of 75 years. Meta-analysis of five studies showed a significantly increased risk of dementia in patients who had hypoglycaemic episodes: pooled odds ratio 1.68 [95% confidence interval (CI) 1.45, 1.95]. We also found a significantly increased risk of hypoglycaemia in patients with dementia: pooled odds ratio from five studies 1.61 (95% CI 1.25, 2.06). Limitations of the study were heterogeneity in the meta-analysis, and uncertain ascertainment of dementia and hypoglycaemic outcomes and temporal relationships. Publication bias may have favoured the reporting of more significant findings.  Conclusions: Our meta-analysis shows a bi-directional relationship between cognitive impairment and hypoglycaemia in older patients. Glucose-lowering therapy should be carefully tailored and monitored in older patients who are susceptible to cognitive decline.